Search

Hi Developers! Welcome to the 17th edition of the InterSystems Ideas bulletin! This time, you can read about the following: ✓ Bringing Ideas to Reality Contest ✓ Recently posted "Community Opportunity" ideas ✓ Vote for implemented ideas you are curious about Bring a Community Opportunity or Future Consideration idea to life and join the Bringing Ideas to Reality Contest! Earn exciting prizes and secure your place in the prestigious Hall of Fame on the Ideas Portal. Here are several recently posted "Community Opportunity" ideas to help you select the one you'd like to implement for the Development Contest: Idea Author Develop productions using YAML @Yuri.Gomes Implement Datadog client that could be easily deployed close to IRIS instance and could be able to send IRIS metrics and logs to Datadog platform (https://www.datadoghq.com). @Mikhail.Khomenko Include an option in HealthShare (UCR) to support natively a vectorized database of patients' clinical data, enabling retrieval-augmented generation (RAG) for communication with market-available large language models (LLMs). @Claudio.Devecchi Know what IRIS version a class, method or property first appeared in @John.Murray Salesforce Interoperability Adapter @Yuri.Gomes IRIS for Health Blockchain connector @Johni.Fisher Last year, we launched a series of articles showcasing ideas brought to life by community members. We’re excited to continue this series and would love your input on which implemented ideas to feature next. Share your opinion by voting in the poll in this announcement: "Which implemented ideas would you like to know more about?" 👏Thank you all for submitting, voting, commenting and implementing existing ideas.👏

Hey Community! We're excited to share the results of the InterSystems Walking Challenge and would like to thank all the members of the Developer Community who joined it and finished the challenge! And now, let's look at the leaderboard 🎊 1 Catarina Meister 18 @Francisco.López1549 2 @Roberta.Scurria 19 @Luciano.Brustia 3 Gabriel Thiébaud 20 @Ina.Haldenmayr 4 Moritz Neunerdt 21 Nicole Sonnek 5 @Benjamin.Spead 22 @Evan.Gabhart 6 Epameinontas Dafnomilis 23 Alesandra Marinova 7 Dimitris Kirikakis 24 @Megumi.Kakechi 8 @Timothy.Leavitt 25 Zhorzheta Palazova 9 @Serval.Cat 26 @LuisAngel.PérezRamos 10 Eduard Gunda 27 @Thembelani.Mlalazi 11 @John.Steiner 28 @Mihoko.Iijima 12 @Iryna.Mykhailova 29 @Oliver.Wilms 13 Kevin Chromik 30 Giorgos Dafnomilis 14 Mykhailo Liubnytskyi 31 Kate Kennedy 15 @Andrii.Mishchenko 32 Oksana Popovichenko 16 @Jason.Neate 33 @jennifer.ames 17 @Daniel.Metcalfe 34 @Hannah.Sullivan 🏅 Every participant who crossed the finish line will receive a medal in recognition of their effort. 🎁 Additionally, the top 30 performers will receive special prizes for their exceptional achievements! NB. Please note that InterSystems employees and contractors are not eligible for these prizes. Congratulations to all finishers! Your medals and prizes will be sent out soon. For that, please provide your preferred shipping address via the app or email marketing.de@intersystems.com. 🎉 Thank you for your participation. We hope you enjoyed the Challenge! Awesome! I'm 12th 🎉 👏 congrats to all involved!! that was a fun challenge :) Congrats Winners!! Looking at the profile flags I was the fastest UK finisher, so I'll take that as a win :)

With InterSystems industry-standard certification exams, you and your team can get certified to validate your skills and demonstrate your expertise in InterSystems technology. Find the right exam for your role! Developers: InterSystems IRIS Development Professional and InterSystems IRIS SQL Specialist System Administrators: InterSystems IRIS System Administration Specialist System Integrators: InterSystems HL7 Interface Specialist CCR Technical Implementers: InterSystems CCR Technical Implementation Specialist Unified Care Record Implementers: HealthShare Unified Care Record Technical Specialist Patient Index implementers: HealthShare Patient Index Technical Specialist Reach out to certification@intersystems.com or visit the InterSystems Certification website for more information.

Hello, In short, I am trying to get the Content-Length of my response, We have a CSP application, when we get a new request we begin to process it, throughout the app we WRITE to the response in different places, now when the response is about to be sent back to the client - we would like to know its Content-Length (in the RESPONSE HEADERS). So it comes down to two questions:1. How can we access the Content-Length of our response? 2. In case we CAN'T - Where in the broker.cls/page/base classes can I observe the moment we actually write the response itself and attach its headers?Thank you! It's not possible because that header it's not handled by IRIS

Hi Developers! Welcome to the 15th edition of the InterSystems Ideas news! We dedicate this news bulletin to: ​​​​✓ Idea Leaders of 2024 ✓ Voting for ideas on Open Exchange ✓ Recently posted ideas waiting to be implemented by the Developer Community In a bit more than half a year, quite a few of Community members have submitted their ideas to the Ideas Portal. We extend our heartfelt thanks to all contributors and want to give a special shout-out to the authors who have shared numerous ideas on the portal this year. Author Number of submitted ideas @Scott.Roth 9 @Veerarajan.Karunanithi9493 8 @LuisAngel.PérezRamos @Victoria.Castillo2990 4 @Stefan.Cronje1399 @Sylvain.Guilbaud @Evgeny.Shvarov Your creativity and dedication are truly inspiring! 👏 You can now vote for ideas that can be implemented by Developer Community members not only on the Ideas Portal but also on the Open Exchange. In the special window (look below for an example screenshot), you can click on the "Vote" button to support the idea. You will see a random Community Opportunity idea whenever you visit Open Exchange. To round up this newsletter, please find the recently posted Community Opportunity ideas Idea Author GUI for Vector DB Management @Ikram.Shah3431 ODBC connection to IRIS Cloud SQL to work from PowerBI and Apache Superset @Vadim.Aniskin Pull in all classes and tables called in a resource at export. @Victoria.Castillo2990 Built in Test data Generator @Veerarajan.Karunanithi9493 Drag and drop web &mobile UI @Veerarajan.Karunanithi9493 Querying from multiple IRIS instances @Veerarajan.Karunanithi9493 Automated Schema Comparison and Synchronization @Veerarajan.Karunanithi9493 Data cleansing tool kit @Veerarajan.Karunanithi9493 Time zone conversion @Veerarajan.Karunanithi9493 IPM (ZPM) extension for VS Code @John.Murray 👏 Many thanks to the authors of these ideas👏 💡 Thank you for reading InterSystems Ideas news. Post your innovative ideas, vote for ideas to support them and implement Community Opportunity ideas to join our Hall of Fame 💡

Hi Developers! Welcome to the 11th edition of the InterSystems Ideas bulletin! This time, you can read about the following: ​​​​✓ Call for ideas concerning Generative AI ✓ New filter on Open Exchange ✓ Recently posted ideas AI is a very important topic nowadays. That's why we have created a separate category on the Ideas Portal for ideas related to this topic. No matter what product or service your idea is about, if it concerns generative AI, put it in the "Generative AI" category. Important: only ideas written by people are accepted, while ideas from AI are not. Open Exchange users can now filter the apps they want to download using the new "Applications from Ideas Portal" filter. You can mention the implemented idea when you upload your new app to OEX. The screenshot below shows what this new filter looks like. To round up this newsletter, here are 25 new ideas that have been posted during the last 2 months. Idea Author Create a WhatsApp Community @Daniel.Aguilar Refresh Button on the small message browse in the Production panel @Stefan.Cronje1399 Ideas contributor of the year @Yuri.Gomes Implement StoreFieldStreamRaw for XML Virtual Documents @LuisAngel.PérezRamos Optimization of the HL7 Schema documentation. @Armin.Gayl Share your recap 2023 on linkedin @Yuri.Gomes Refactoring in VsCode @Pierre.LaFay2520 Train a Large Language Model for a Development AI like GitHub Copilot on ISC @Stefan.Cronje1399 Best practices in a documentation chapter @Pierre.LaFay2520 Add access to Ideas Portal from Global Masters @LuisAngel.PérezRamos Add info about the versions in which the functionality works in the documentation @LuisAngel.PérezRamos Ranking - top 10 - Open Exchange @Yuri.Gomes Date format in LOAD DATA to indicate a date/datetime format other than 'yyyy-mm-dd hh:mm:ss' @Sylvain.Guilbaud cheat sheets @Pierre.LaFay2520 Add JS (JSON) date format to formats used in $ZDATE & $ZDATEH @Pierre.LaFay2520 IRIS Cloud SQL Integration with AWS Secrets Manager @Stefan.Cronje1399 Web Interface to convert HL7 V2 to FHIR | InterSystems Ideas @Muhammad.Waseem Implement support for FHIRPath Patch resources on InterSystems FHIR Server @Maksym.Shcherban AI extensibility Prompt keyword for Class and Method implementation. Also Prompt macro generator. @Alex.Woodhead BPL, DTL and Rulesets in JSON and/or YAML @Stefan.Cronje1399 Create List and Array Data Types that uses a PPG and not memory @Stefan.Cronje1399 To integrate InterSystems IRIS for Health with Open Source Drools @William.Liu Mirroring : add automatic synchronization between two sync-members of non-mirrored settings @Sylvain.Guilbaud Implement shopping cart to get multiple prizes from Global Masters at one time @LuisAngel.PérezRamos add AI to Instance Monitor to prevent major incident @Pietro.Montorfano 👏 Authors, thank you for posting these ideas 👏 🙏 Don't forget to advertise your idea and other ideas you like to your peers and all Developer Community members. The more votes your idea has, the larger the chance of its promotion to the next status. 🚀

Hi Community, We are pleased to invite you to the next InterSystems online programming contest, which is focused on Python! 🏆 InterSystems Python Contest 🏆 Duration: July 15 - August 4, 2024 Prize pool: $14,000 The topic Develop any solution using InterSystems IRIS, InterSystems IRIS for Health, or IRIS Cloud SQL that uses Python (or Embedded Python) as a programming language. General Requirements: An application or library must be fully functional. It should not be an import or a direct interface for an already existing library in another language (except for C++, where you really need to do a lot of work to create an interface for IRIS). It should not be a copy-paste of an existing application or library. Accepted applications: new to Open Exchange apps or existing ones, but with a significant improvement. Our team will review all applications before approving them for the contest. The application should work either on IRIS, IRIS for Health or IRIS Cloud SQL. The first two could be downloaded as host (Mac, Windows) versions from Evaluation site, or can be used in the form of containers pulled from InterSystems Container Registry or Community Containers: intersystemsdc/iris-community:latest or intersystemsdc/irishealth-community:latest . The application should be Open Source and published on GitHub. The README file to the application should be in English, contain the installation steps, and either the video demo or/and a description of how the application works. No more than 3 submissions from one developer are allowed. NB. Our experts will have the final say in whether the application is approved for the contest or not based on the criteria of complexity and usefulness. Their decision is final and not subject to appeal. Prizes 1. Experts Nomination - a specially selected jury will determine winners: 🥇 1st place - $5,000 🥈 2nd place - $3,000 🥉 3rd place - $1,500 🏅 4th place - $750 🏅 5th place - $500 🌟 6-10th places - $100 2. Community winners - applications that will receive the most votes in total: 🥇 1st place - $1,000 🥈 2nd place - $750 🥉 3rd place - $500 🏅 4th place - $300 🏅 5th place - $200 If several participants score the same amount of votes, they all are considered winners, and the money prize is shared among the winners. Who can participate? Any Developer Community member, except for InterSystems employees (ISC contractors allowed). Create an account! Developers can team up to create a collaborative application. 2 to 5 developers are allowed in one team. Do not forget to highlight your team members in the README of your application – DC user profiles. Important Deadlines: 🛠 Application development and registration phase: July 15, 2024 (00:00 EST): Contest begins. July 28, 2024 (23:59 EST): Deadline for submissions. ✅ Voting period: July 29, 2024 (00:00 EST): Voting begins. August 4, 2024 (23:59 EST): Voting ends. Note: Developers can improve their apps throughout the entire registration and voting period. Helpful Resources: ✓ Documentation: Embedded Python Documentation Native SDK for Python Documentation PEX Documentation ✓ Example applications and libraries: iris-fastapi-template iris-django-template iris-flask-template interoperability-python pex-demo python-examples Python Faker ✓ Online courses: Learning Path Writing Python Application with InterSystems ✓ Videos: Introduction to Embedded Python Embedded Python: Bring the Python Ecosystem to Your ObjectScript App Embedded Python for ObjectScript Developers: Working with Python and ObjectScript Side-By-Side Embedded Python with Interoperability InterSystems IRIS Native Python API in AWS Lambda ✓ For beginners with IRIS: Build a Server-Side Application with InterSystems IRIS Learning Path for beginners ✓ For beginners with ObjectScript Package Manager (IPM): How to Build, Test and Publish IPM Package with REST Application for InterSystems IRIS Package First Development Approach with InterSystems IRIS and IPM ✓ How to submit your app to the contest: How to publish an application on Open Exchange How to submit an application for the contest Need Help? Join the contest channel on InterSystems' Discord server or talk with us in the comment to this post. We're waiting for YOUR project – join our coding marathon to win! By participating in this contest, you agree to the competition terms laid out here. Please read them carefully before proceeding. + the IRIS 2024.1 release notes : Python BPL Editor WSGI Web Apps That sounds great!!! Excited about it  A "Must read" article for using Python with IRIS. Hi Developers! The "Kick-off Webinar for InterSystems Python Contest" recording is on the InterSystems Developers YouTube channel! 🔥Enjoy! ⏯️Kick-off Webinar for InterSystems Python Contest Hey Devs! There are only 4 days left before the end of the registration phase! Upload your applications and join the contest! One app has been added to the contest, check it out! iris-email-analyzer-app by @Eric.Mariasis Developers! Three more articles have been added to the contest! sheltershare by @Zeljko.Sucic ServiceInspection by @Wolis.OliavrIRIS RAG App by @Alex.Alcivar There are only two days left to join the contest! We are waiting for you!

InterSystems’ team is heading to the HackMIT 2024 hackathon, taking place on September 14-15, 2024! HackMIT is a weekend-long event where students from around the globe work together to design and build innovative technology projects. Students will focus on four main areas for innovation: healthcare, sustainability, education, and interactive media. Interested to become a mentor? Send a direct message to us!

Hi Community! Welcome to the 12th edition of the InterSystems Ideas news! Here's what you can expect from it: ​​​​✓ Call for ideas about InterSystems IRIS Cloud SQL ✓ Get inspired by ideas from the InterSystems Ideas portal ✓ Recently implemented ideas Many Developer Community members have already tested and used InterSystems' new service, InterSystems IRIS Cloud SQL. You must have bright ideas related to this service, and we are waiting for them in the new Ideas Portal category dedicated to this database-as-a-service (DBaaS). If you are on the lookout for an interesting idea to implement or just want inspiration to create a new application and publish it on the Open Exchange, you can now get to the Ideas Portal directly from the "Applications" item of the Open Exchange main menu. Since the launch of the Ideas Portal, 54 ideas related to the InterSystems products and resources for developers have already been implemented. And 12 of them were realized in the first quarter of 2024: Idea Implementation of the idea (project) Developer Add access to Ideas Portal from Global Masters by @LuisAngel.PérezRamos see Global Masters interface InterSystems Make a module which will let to use IRIS and Firebase framework by @Evgeny.Shvarov irisfirebase @Daniel.Aguilar Ideas contributor of the year by @Yuri.Gomes Top InterSystems Ideas Contributors for 2023 InterSystems Web Interface to convert HL7 V2 to FHIR by @Muhammad.Waseem iris-fhir-lab @Muhammad.Waseem Expose Global Masters Badges by @henry Developer Community Release, January 2024 InterSystems On Developer Community put animated GIFs behind a play/pause button by @John.Murray Developer Community Release, January 2024 InterSystems Tutorial about custom login with class extending %CSP.Login by @Yuri.Gomes custom-login @Yuri.Gomes Add a Support category to the Services offered. by @Paul.Hurley see Partner Directory interface InterSystems Sync Community and Open Exchange users by @LuisAngel.PérezRamos see Community and Open Exchange interfaces InterSystems Have nicknames for community users by @Minoru.Horita Developer Community Release, January 2024 InterSystems Introduce an interoperability adapters to import/export data from Google Docs and Google Sheets by @Evgeny.Shvarov googlesheets-adapter @Nikolay.Soloviev Tribute banner and interview for GM million pointers by @Yuri.Gomes Celebrating a True Pillar of the Developer Community: A Journey of Dedication and Expertise @Robert.Cemper1003 and InterSystems 👏 Thank you for implementing and posting these ideas 👏 Stay tuned for our next announcements! In the meantime, post your brilliant ideas, and advertise ideas you like to your peers and all Developer Community members!

Hi Developers! Welcome to the 14th edition of the InterSystems Ideas news!! This time, you can read about the following: ​​​​✓ Hints & Tips for the participants of the 3rd InterSystems Ideas Contest ✓ Brand new Ideas Portal landing page ✓ Recently implemented ideas After only 10 days, we already have 14 accepted ideas. We thank all Community members who posted their ideas for the Contest. Special thanks to the @Victoria Castillo for posting 4 brilliant ideas! 👏 Make a difference by posting ideas focused on solving real-world problems. Share your creativity with all Community Members and InterSystems team! Some other ideas were posted during the Contest period but were not accepted because they: contained AI generated text were the same as the existing ideas on the Portal were related to the InterSystems TrakCare product were too general and non-specific Please read the requirements for ideas in the Ideas Contest announcement carefully to ensure your idea is accepted. If you are new to the Ideas Portal, you can read "Ideas promotion rules" on the Portal Guide page and an article dedicated to the creation of new ideas. We hope these hints will help you post your ideas for the Ideas Contest. The Ideas Portal landing page was launched recently. If you are new to the Ideas Portal, please look at it to understand the logic of how this resource works and find useful links. If you are an experienced ideas author, please share your feedback on how we can improve it. To round up this newsletter, please find below the ideas implemented during the second quarter of 2024. Idea Implementation of the idea (project) Developer Assert CodeAssist prompting by @Alex.Woodhead starting from InterSystems IRIS 2023.3.0 InterSystems Introduce WITH into IRIS SQL engine by @Evgeny.Shvarov starting from InterSystems IRIS 2024.1 InterSystems Implement support for FHIRPath Patch resources on InterSystems FHIR Server by @Maksym.Shcherban see InterSystems FHIR Server documentation InterSystems Reference architectures for VIPs on SDNs in GCP by @Eduard.Lebedyuk VIP in GCP @Mikhail.Khomenko ​​​​@Eduard.Lebedyuk InterSystems EF Core Provider by @Evgeny.Shvarov InterSystems 👏 Many thanks to the implementors and authors of these ideas👏 💡Post your ideas till the 30th of June to participate in the 3rd InterSystems Ideas Contest, vote and comment on ideas participating in the Contest! 💡

InterSystems’ team is heading to the IC HACK ‘25 hackathon, taking place on February 1-2, 2025 in London, UK. IC HACK ‘25 is an annual hackathon organised by Imperial College London Department of Computing Society. It is the biggest student-run hackathon in Europe, where over 700 of the UK's most creative and talented students work together for 24 hours of learning, building, fun, and networking.

InterSystems’ team is heading to the TreeHacks 2025 hackathon, taking place on February 14-16, 2025! TreeHacks 2025 is Stanford's premier hackathon. The country's brightest engineering students are flown to Stanford's campus to build solutions to the world's largest challenges for 36 hours straight. Students will focus on four main areas for innovation: healthcare, sustainability, education, fintech & security, and entertainment. Interested to become a mentor? Send a direct message to us!

In 2023, according to IDC, Salesforce's market share in CRM reached 21.7%. This company owns a substantial amount of critical corporate business processes and data, so the InterSystems IRIS must have an interoperability connector to fetch data from the Salesforce data catalog. This article will show you how to get any data hosted by Salesforce and create an interoperation production to get data and send it to such targets as files and relational databases. Create your development account on Salesforce: 1. Go to https://developer.salesforce.com/signup: 2. Fill out the form and click "Sign Me Up" to create your DEV account.3. To explore the Salesforce data catalog, log in and proceed to https://[YOUR-INSTANCE]-dev-ed.develop.lightning.force.com/lightning/setup/ObjectManager/home (my case can be found via https://ymservices-dev-ed.develop.lightning.force.com/lightning/setup/ObjectManager/home). 4. Click the Account Label to go to the Account data object details (the API Name is Account): 5. Head to the Fields & Relationships tab to see the Account fields: 6. We will consume the Account data in our sample. Configure the OAuth credentials: To consume the Salesforce API, you must configure the OAuth credentials:1. Go to https://[YOUR SALESFORCE INSTANCE].lightning.force.com/lightning/setup/NavigationMenus/home.2. Click the button "New Connected App": 3. Select "Create a Connected App" and click "Continue": 4. Fill in the form according to the options shown in the image (use your email on Contact Email) and click "Save": 5. Wait for 30 to 45 minutes and click "Continue". Then click "Manage Consumer Details": 6. You must validate your identity (check your email to get the verification code): 7. Copy the consumer key and consumer secret since you will need them later.8. Now you should edit a few OAuth policies.9. Proceed to https://[your salesforce instance].lightning.force.com/lightning/setup/ConnectedApplication/home and click "Edit" next to the IRISInterop connected application. 10. In the OAuth Policies section, pick the following options and click the "Save" button at the bottom:a. Permitted Users: All users may self-authorizeb. IP Relaxation: Relax IP restrictions 11. At this point, we are ready to consume the Salesforce APIs. Test the consumption of the Salesforce API on Postman: 1. If you do not have Postman, you can download it at https://www.postman.com/downloads/.2. Start your Postman and test the login API to get a new token.3. Create a POST call as indicated below:a. grant_type: passwordb. client_id: your consumer keyc. client_secret: your consumer secretd. username: your username (the one you utilize to access your Salesforce instance web interface)e. password: your user password (the one you employ to access your Salesforce instance web interface) 4. Copy the access_token to use it on the Authorization header for all the upcoming Salesforce API calls5. Now, we are ready to consume the Salesforce Query API to get the current leads. Set your call as shown below:a. URL: https://[your salesforce instance].my.salesforce.com/services/data/v58.0/query?q=SELECT+Id,+Name+FROM+Leadb. On headers, add the Header Authorization with the value: Bearer [the access_token returned from login service] 6. Check out the records with the data about leads.7. All set! It is time to create an interoperability adapter with production to allow us to integrate Salesforce data with any other system, service, API, or data repository. About Salesforce API We will employ the following API to interoperate with Salesforce: Authentication API to get Authorization token: curl --location 'https://ymservices-dev-ed.develop.my.salesforce.com/services/oauth2/token' \ --form 'grant_type="password"' \ --form 'client_id="3MVG91oqviqJK………….fT_0CZ9M.80M_THhxS51LKc7IHk2c5qQt63wD_"' \ --form 'client_secret="6B8860E……………..B3AC840E568D44837094ADC"' \ --form 'username="yurimarx@gmail.com"' \ --form 'password="**************"' Query API (this API finds records using the Salesforce Object Query Language (SOQL) which is similar to SQL language): SELECT one or more fields FROM an object WHERE filter statements and, optionally, results are ordered Example: SELECT Id, Name FROM Account WHERE Name = 'Sandy' A cURL sample: curl --location --request GET 'https://ymservices-dev-ed.develop.my.salesforce.com/services/data/v62.0/query?q=SELECT+Id%2C+Name+FROM+Lead' \ --header 'Authorization: Bearer 00Dbm00000CTkBv!AQEAQFIMdVSCElZDlb0axaaWBGm5Hacr6EN.77fgNvc4zpq4N22Hgf1THQydrewNQNM5W9cttWKyOtZMfDYTHTS4FqlS2uov' For more details about the language capabilities, check out the link below:https://developer.salesforce.com/docs/atlas.en-us.soql_sosl.meta/soql_sosl/sforce_api_calls_soql_select.htm. sObject API (this API executes CRUD operations and describes the Salesforce objects and their metadata). The URL path is /services/data/salesforce version/sobjects/salesforce object/salesforce record ID (not to create). Please note that you must set the Authorization header with the login token. Examine the samples below: 1. Create record: curl https://MyDomainName.my.salesforce.com/services/data/v63.0/sobjects/Account/ -H "Authorization: Bearer token" -H "Content-Type: application/json" -d "@newaccount.json".2. Update record: curl https://MyDomainName.my.salesforce.com/services/data/v63.0/sobjects/Account/001D000000INjVe -H "Authorization: Bearer token" -H "Content-Type: application/json" -d @patchaccount.json -X PATCH3. Delete record: curl https://MyDomainName.my.salesforce.com/services/data/v63.0/sobjects/Account/001D000000INjVe -H "Authorization: Bearer token" -X DELETE4. Get by Id: curl https://MyDomainName.my.salesforce.com/services/data/v63.0/sobjects/Account/001D000000INjVe -H "Authorization: Bearer token" -X GET5. Get metadata information: curl https://MyDomainName.my.salesforce.com/services/data/v63.0/sobjects/Account/ -H "Authorization: Bearer token"More information about sObject API is available via the next link: https://developer.salesforce.com/docs/atlas.en-us.api_rest.meta/api_rest/intro_rest.htm. Creating the interoperability adapter and production Get the sample application: To illustrate how to interoperate with Salesforce, first, you should get my sample app the iris-salesforce application, on Open Exchange via the link https://openexchange.intersystems.com/package/Iris-Salesforce. Download or clone the application from GitHub to see the implementation details described in the upcoming sections. Creating the Interoperability Adapter for Salesforce Since there is no official interoperability adapter for Salesforce on InterSystems IRIS, we need to create one: Class dc.irissalesforce.SalesforceOutboundAdapter Extends Ens.OutboundAdapter { Property SFVersion As %String(VALUELIST = ",v62.0,v61.0,v60.0,v59.0,v58.0,v57.0,v56.0,v55.0,v54.0,v53.0,v52.0,v51.0,v50.0,v49.0,v48.0,v47.0,v46.0,v45.0,v44.0,v43.0,v42.0,v41.0,v40.0,v39.0,v38.0,v37.0,v36.0,v35.0,v34.0,v33.0,v32.0,v31.0") [ InitialExpression = "v62.0" ]; Property SFCredsFile As %String(MAXLEN = 1000); Property SSLConfig As %String; Parameter SETTINGS = "SFVersion:Connect,SFCredsFile:Connect:fileSelector,SSLConfig:Connect:sslConfigSelector"; Method PrepareRequest(sfToken As %DynamicObject) As %Net.HttpRequest { Set sfReq = ##class(%Net.HttpRequest).%New() Set sfReq.Server = sfToken.instance Set sfReq.Https = 1 Set sfReq.SSLConfiguration = ..SSLConfig Do sfReq.SetHeader("Authorization", "Bearer "_sfToken."access_token") Return sfReq } /// Query Salesforce with SQL query to get a JSON response with business data Method Query(queryRequest As dc.irissalesforce.QueryMessage, Output queryResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set sfToken = {} Set queryResponse = ##class(Ens.StreamContainer).%New() Try { Do ..GetAuthotizationToken(.sfToken) Set sfReq = ..PrepareRequest(sfToken) Do sfReq.SetParam("q", queryRequest.Query) Do sfReq.Get("/services/data/"_..SFVersion_"/query") Do queryResponse.StreamSet(sfReq.HttpResponse.Data) } Catch ex { Set sc = ex.AsStatus() } Return sc } /// Get Salesforce data with ID to get a JSON response with business data Method GetById(idRequest As dc.irissalesforce.ByIdMessage, Output queryResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set sfToken = {} Set queryResponse = ##class(Ens.StreamContainer).%New() Try { Do ..GetAuthotizationToken(.sfToken) Set sfReq = ..PrepareRequest(sfToken) Do sfReq.Get("/services/data/"_..SFVersion_"/sobjects/"_idRequest.ClassName_"/"_idRequest.Id) Do queryResponse.StreamSet(sfReq.HttpResponse.Data) } Catch ex { Set sc = ex.AsStatus() } Return sc } /// Create a new Salesforce object with json data Method Create(createRequest As dc.irissalesforce.CreateMessage, Output createResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set sfToken = {} Set createResponse = ##class(Ens.StreamContainer).%New() Try { Do ..GetAuthotizationToken(.sfToken) Set sfReq = ..PrepareRequest(sfToken) Set sfReq.ContentType = "application/json" Do sfReq.EntityBody.Write(createRequest.Content) Do sfReq.Post("/services/data/"_..SFVersion_"/sobjects/"_createRequest.ClassName_"/") Do createResponse.StreamSet(sfReq.HttpResponse.Data) } Catch ex { Set sc = ex.AsStatus() } Return sc } /// Update the Salesforce object with json data Method Update(updateRequest As dc.irissalesforce.UpdateMessage, Output updateResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set sfToken = {} Set updateResponse = ##class(Ens.StreamContainer).%New() Try { Do ..GetAuthotizationToken(.sfToken) Set sfReq = ..PrepareRequest(sfToken) Set sfReq.ContentType = "application/json" Do sfReq.EntityBody.Write(updateRequest.Content) Do sfReq.Patch("/services/data/"_..SFVersion_"/sobjects/"_updateRequest.ClassName_"/"_updateRequest.Id_"/") Do sfReq.HttpResponse.OutputToDevice() Do updateResponse.StreamSet(sfReq.HttpResponse.Data) } Catch ex { Set sc = ex.AsStatus() } Return sc } /// Delete the Salesforce object by id Method Delete(deleteRequest As dc.irissalesforce.DeleteMessage, Output deleteResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set sfToken = {} Set deleteResponse = ##class(Ens.StreamContainer).%New() Try { Do ..GetAuthotizationToken(.sfToken) Set sfReq = ..PrepareRequest(sfToken) Set sfReq.ContentType = "application/json" Do sfReq.Delete("/services/data/"_..SFVersion_"/sobjects/"_deleteRequest.ClassName_"/"_deleteRequest.Id_"/") Do sfReq.HttpResponse.OutputToDevice() Do deleteResponse.StreamSet(sfReq.HttpResponse.Data) } Catch ex { Set sc = ex.AsStatus() } Return sc } /// Get OAuth token Method GetAuthotizationToken(Output oauthToken As %DynamicObject) As %Status { Set sc = $$$OK Set sfCreds = {} Try { Set sfCreds = {}.%FromJSONFile(..SFCredsFile) Set sfReq = ##class(%Net.HttpRequest).%New() Set sfReq.Server = sfCreds.instance Set sfReq.Https = 1 Set sfReq.SSLConfiguration = ..SSLConfig Do sfReq.InsertFormData("grant_type", "password") Do sfReq.InsertFormData("client_id", sfCreds.clientId) Do sfReq.InsertFormData("client_secret", sfCreds.clientSecret) Do sfReq.InsertFormData("username", sfCreds.username) Do sfReq.InsertFormData("password", sfCreds.password) Do sfReq.Post("/services/oauth2/token") Set oauthToken = {}.%FromJSON(sfReq.HttpResponse.Data.Read()) Set oauthToken.instance = sfCreds.instance } Catch ex { Set sc = ex.AsStatus() } Return sc } } Explore the notes about this implementation:1. The class extends Ens.OutboundAdapter because Salesforce is not used as the first step of the interoperability production (it does not initiate the integration process since it is output, not input).2. Developers utilize the SF Version to choose the Salesforce version.3. The SFCredsFile sets the credentials file with sensitive information to connect to Salesforce (instance name, username, password, client id, and client secret).4. SSLConfig sets the SSL configuration to call the Salesforce HTTPS endpoint.5. SETTINGS are used by IRIS to demonstrate the developer input fields for set values to the SF Version, SFCredsFile, and SSLConfig properties.6. The PrepareRequest method creates the HTTPRequest object and establishes the common properties for all other methods.7. The Query method calls the Query API with the Select sentence passed on QueryMessage.8. The GetById method calls the sObject API passing the Salesforce Object class name and the unique Id required to get the record data.9. The Create method calls the sObject API passing the Salesforce Object class name and the data in JSON format to save a new record with the business data.10. The Update method calls the sObject API passing the Salesforce Object class name, the unique Id, and the data that will be altered.11. The Delete method calls the sObject API passing the Salesforce Object class name and the unique Id needed to delete the record.12. The GetAuthorizationToken method calls the oauth2 API passing the credentials data to get the authorization token needed to call all the other APIs.13. The adapter methods return Ens.StreamContainer to support any content size. Creating the Interoperability Operation for Salesforce Since InterSystems IRIS does not have an official interoperability operation for Salesforce, we must build one with the help of the created adapter: Class dc.irissalesforce.SalesforceOperation Extends Ens.BusinessOperation { Parameter ADAPTER = "dc.irissalesforce.SalesforceOutboundAdapter"; Parameter INVOCATION = "Queue"; /// Get Salesforce data from SQL query Method Query(queryRequest As dc.irissalesforce.QueryMessage, Output queryResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set queryResponse = ##class(Ens.StreamContainer).%New() Set sc = ..Adapter.Query(queryRequest, .queryResponse) Return sc } /// Get Salesforce object data from Object ID Method GetById(idRequest As dc.irissalesforce.ByIdMessage, Output queryResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set queryResponse = ##class(Ens.StreamContainer).%New() Set sc = ..Adapter.GetById(idRequest, .queryResponse) Return sc } /// Create Salesforce object with data from JSON Method Create(createRequest As dc.irissalesforce.CreateMessage, Output createResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set createResponse = ##class(Ens.StreamContainer).%New() Set sc = ..Adapter.Create(createRequest, .createResponse) Return sc } /// Update Salesforce object with data from JSON Method Update(updateRequest As dc.irissalesforce.UpdateMessage, Output updateResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set updateResponse = ##class(Ens.StreamContainer).%New() Set sc = ..Adapter.Update(updateRequest, .updateResponse) Return sc } /// Delete Salesforce object by Id Method Delete(deleteRequest As dc.irissalesforce.DeleteMessage, Output deleteResponse As Ens.StreamContainer) As %Status { Set sc = $$$OK Set deleteResponse = ##class(Ens.StreamContainer).%New() Set sc = ..Adapter.Delete(deleteRequest, .deleteResponse) Return sc } XData MessageMap { <MapItems> <MapItem MessageType="dc.irissalesforce.QueryMessage"> <Method>Query</Method> </MapItem> <MapItem MessageType="dc.irissalesforce.ByIdMessage"> <Method>GetById</Method> </MapItem> <MapItem MessageType="dc.irissalesforce.CreateMessage"> <Method>Create</Method> </MapItem> <MapItem MessageType="dc.irissalesforce.UpdateMessage"> <Method>Update</Method> </MapItem> <MapItem MessageType="dc.irissalesforce.DeleteMessage"> <Method>Delete</Method> </MapItem> </MapItems> } } Examine the notes about this implementation: 1. The class extends the Ens.BusinessOperation (the base class for producing custom business operations).2. The ADAPTER parameter is dc.irissalesforce.SalesforceOutboundAdapter. It is your custom Adapter with methods to integrate with Salesforce.3. The INVOCATION parameter with Queue value means the message is created within one background job and placed on a queue simultaneously releasing the original job.4. The methods Query, GetById, Create, Update, and Delete call the respective methods on the Adapter class (dc.irissalesforce.SalesforceOutboundAdapter) utilizing..Adapter.[name of method].5. Each method has a specific message class:a. Method Query: QueryMessage class with Query property used to execute the SQL sentence.b. Method GetById: GetByIdMessage class with the property Id with the unique id value and ClassName with the object class name to be returned.c. Method Create: CreateMessage class with the property ClassName for object class name to be created and the Content property with JSON business data to be saved.d. Method Update: UpdateMessage class with the property ClassName for object class name to be made, the Id property to set the record to be updated and the Content property with JSON business data to be saved.e. Method Delete: DeleteMessage class with the property ClassName for object class name and the Id property for the record to be deleted.6. The XData MessageMap maps the methods to be called based on the MessageType sent to the Operation. For instance, if the Salesforce Operation receives a CreateMessage, the Create method will be called.7. The message classes extend %Persistent to preserve the message history and Ens.Util.RequestBodyMethods to monitor, audit, and trace messages better. Creating a Production to use the new Operation Finally, to test our Salesforce business operation, we will build a new Production:1. Check out the port used by IRIS (in my case, it is 51925): 2. Go to the Management Portal http://localhost:51926/csp/sys/UtilHome.csp?$NAMESPACE=IRISAPP3. Proceed to Interoperability > Configure > Production: 4. Click the New button: 5. Set the values for the new Production and click OK: 6. Move to Settings tab > Development and Debugging section, check the Testing Enabled option, and click the Apply button to save your selection: 7. Click the Plus button located near Operations: 8. Set the values to the new operation (Operation Class is our SalesforceOperation class): 9. Well done! We have TestOperation ready for our tests: 10. Select TestOperation, go to Settings tab > Connect section, and establish the following values:a. SFVersion: v62.0b. SFCredsFile: path for credentials file, (in my case, it is /usr/irissys/creds.json)c. SSL Configuration: with the SSL configuration, (in my case, it is pm.community.intersystems.com)11. Click the Apply button to save the TestOperation configuration: 12. Click the production Start button to execute the operation: 13. At this point, the Operation color should turn dark green: 14. Pick the TestOperation operation and proceed to the Actions tab. Then click the button Test: 15. Set the values to test Create:Request Type: dc.irissalesforce.CreateMessageContent: { "LastName": "InterSystems", "FirstName": "IRIS", "Salutation": "Mr.", "Title": "Datafabric sales", "Company": "InterSystems", "MobilePhone": "555555", "Email": "test@gmail.com", "LeadSource": "Web", "Status": "Working - Contacted", "Rating": "Warm" } ClassName: Lead 16. Click the button Invoke Testing Service and check out the results: 17. Click the link Visual Trace to see the details of the tests: 18. Click the [3] step and move to the Contents tab to explore the result details: 19. The record was created successfully, so copy the ID (in my case, it is 00Qbm00000B4YrDEAV).20. Close the Visual Trace, and choose the Request Type dc.irissalesforce.ByIdMessage. Then expand the section Request Details: 21. Set the values ID with the ID you copied and ClassName with Lead: 22. Click the button Invoke Testing Service and study the results: 23. Click the Visual Trace, and pick [3] step. It will display the results on the Contents tab: 24. At this point, you should see the JSON content that you input in the previous steps.25. Test the remaining message types and enjoy!

Hi, Community! Looking to get actionable insights from your healthcare research? See how InterSystems OMOP can help: 👨‍🔬What is InterSystems OMOP? With InterSystems OMOP—a cloud-based software-as-a-service—you can transform clinical data into the OMOP format and get faster insights. Benefits include: Create research data repositories efficiently. Easily ingest, transform, and store data. 🎬Watch the video to learn more!

As we all know, InterSystems is a great company. Their products can be just as useful as they are complex. Yet, our pride sometimes prevents us from admitting that we might not understand some concepts or products that InterSystems offers for us. Today we are beginning a series of articles explaining how some of the intricate InterSystems products work, obviously simply and clearly. In this essay, I will clarify what Machine Learning is and how to take advantage of it.... because this time, you WILL KNOW for sure what I am talking about. What (the hell) is Machine Learning? Machine Learning is a branch of artificial intelligence that focuses on developing algorithms and models that enable computers to learn to perform specific tasks based on data, without being explicitly programmed for each task. Instead of following specific instructions, machines learn through experience, identifying patterns in data and making predictions or decisions based on them. The process involves feeding algorithms with datasets (called training sets) to make them learn and improve their performance over time. Those algorithms can be designed to perform a wide range of tasks, including image recognition, natural language processing, financial trend prediction, medical diagnosis, and much more. In summary, Machine Learning allows computers to learn from data and improve with experience, enabling them to perform complex tasks without explicit programming for each situation autonomously... It is a lovely definition. Yet, I guess you need an example, so here we go: Well, imagine that every day you write down somewhere the time of sunrise and sunset. If somebody asked you whether the sun would rise the next day, what would you say? All you have noticed was only the time of sunrise and sunset.. By observing your data, you would conclude that with 100% probability, the sun will rise tomorrow. However, you cannot ignore the fact that there is a chance that, due to a natural catastrophe, you will not be able to see the sun rising the next day. That is why you should say that the likelihood of witnessing a sunrise the following day is, in fact, 99.99%. Considering your personal experience, you can provide an answer that matches your data. Machine Learning is the same thing but done by a computer.. Look at the table below: A B 1 2 2 4 3 6 4 8 How do columns A and B relate to each other? The answer is easy, the value of B is double the A. B=A*2, is a pattern. Now, examine the other table: A B 1 5 2 7 3 9 4 11 This one is a bit more complicated…. If you haven't uncovered the pattern, it is B=(A*2) +3. A human, for instance, can deduce the formula, meaning that the more data you have, the easier it is to guess the pattern behind this mystery. So, Machine Learning uses the same logic to reveal the pattern hidden in the data. How to start? First, you will need a computer, Yes, since this article is about Machine Learning, having only a notebook and a pencil will not be enough. Second, you will require an instance of IRIS Community. You can download a Docker image and execute your test here. Note, that it must have ML integrated, e.g., the latest version of InterSystems IRIS Community: docker pull intersystems/iris-ml-community:latest-em or docker pull intersystems/iris-community:latest If you need another platform, check https://hub.docker.com/r/intersystems/iris-ml-community/tags or https://hub.docker.com/r/intersystems/iris-community/tags. Then, create a container from this container and run it: docker run --name iris-ml -d --publish 1972:1972 --publish 52773:52773 intersystems/iris-m If you are "old-school", you can download a free version for evaluation. Yet, it is important to have an InterSystems account. Check it out at https://login.intersystems.com/login/SSO.UI.Register.cls. Afterward, ask for an evaluation copy at https://evaluation.intersystems.com/Eval/. Install it and run your instance. Now, access the IRIS portal. http://localhost:52773/csp/user/EnsPortal.ProductionConfig.zen User: Superuser Pass: SYS Note: You might be asked to change the password the first time. Do not be afraid, just come up with a password that you can easily remember. Open the "Machine learning configuration" to review the versions you installed. At this point, you can see the provider configurations of ML installed. Earth, "water" and fire... what is the best? All of them are good. The important thing is how to train your dragon, I mean... your data. Explore more info about the existing models: AutoML: AutoML is an automated Machine Learning system developed by InterSystems and housed within the InterSystems IRIS® data platform. It is designed to build accurate predictive models quickly using your data. It automates several key components of the machine-learning process. Click the link below to see more info: https://docs.intersystems.com/iris20241/csp/docbook/Doc.View.cls?KEY=GAUTOML_Intro H2O: It is an open-source Machine Learning model. The H2O provider does not support the creation of time series models. Follow the link below to discover more: https://docs.intersystems.com/irislatest/csp/docbook/DocBook.UI.Page.cls?KEY=GIML_Configuration_Providers#GIML_Configuration_Providers_H2O PMML: (Predictive Modelling Markup Language). It is an XML-based standard that expresses analytics models. It provides a way for applications to define statistical and data mining models so that they can be easily reused and shared. Check out the link below for more info: https://docs.intersystems.com/irislatest/csp/docbook/DocBook.UI.Page.cls?KEY=APMML What is the first step? Just like in the sunset and sunrise example, we need some data to train our model. It is essential to know the data objective and the values that should be predicted. It is also critical to have clear data without any duplicates. You must find out what the minimum set of data is as well. I am going to use the AutoML provider because it is from Intersystems, ha-ha 😉 There are a few kinds of algorithms: Decision trees: First, the information is classified, then the next question is applied to evaluate the probability. Example: Will it rain tomorrow? Check if the sky is cloudy (very or slightly) or clear. If it is very cloudy, then check the humidity. After that, check the temperature... If it is very cloudy, with high humidity and low temperature, then it will rain tomorrow.. Random forests: It is a set of decision trees, each of which "votes" for a class. The majority of the votes define the selected model. Neural networks: It does not mean that Skynet is coming... However, it is too complicated to explain in just a few words. The general idea is to "copy" the function of human neurons. It means that each input data gets analyzed by a "neuron", which, in turn, provides the input data to the next "neuron" to analyze the output data. If you wish to play around with neural networks using Python, you can create one and check how it works. Please, have a look at https://colab.research.google.com/drive/1XJ-Lph5auvoK1M4kcHZvkikOqZlmbytI?usp=sharing. Through the link above, you can run a routine in Python, with the help of the TensorFlow library. To get the pattern of tables A and B do the following: import tensorflow as tf import numpy as np tableA = np.array([1, 2, 3, 4, 5, 6, 7], dtype=float) tableB = np.array([5, 7, 9, 11, 13, 15, 17], dtype=float) hidden1 = tf.keras.layers.Dense(units=3, input_shape=[1]) hidden2 = tf.keras.layers.Dense(units=3) exit = tf.keras.layers.Dense(units=1) model = tf.keras.Sequential([hidden1, hidden2, exit]) model.compile( optimizer=tf.keras.optimizers.Adam(0.1), loss='mean_squared_error' ) print("Start training...") history = model.fit(tableA, tableB, epochs=1000, verbose=False) print("Model trained!") import matplotlib.pyplot as plt plt.xlabel("# Epoch") plt.ylabel("Loss magnitud") plt.plot(history.history["loss"]) print("Doing a predicction!") result = model.predict([100]) print("The result is " + str(result) ) print("Internal variables of the model") print(hidden1.get_weights()) print(hidden2.get_weights()) print(exit.get_weights()) The code above utilizes the values of A and B to create a model to compare and discover the relation between both values. When I have done the prediction, it retrieves the correct value, in this sample the prediction is 203. How does it work in IRIS? Machine Learning in IRIS is called “integratedML”. It has been implemented since InterSystems IRIS 2023.2 as an Experimental Feature, meaning that it is not supported for production environments. However, the feature is well-tested, and InterSystems believes it can add significant value to customers. You can find more information in Using integratedML documentation. Still, since this is an ML lesson for beginners, I will explain how to operate it as simply as possible. Note: I am utilizing a docker with an image from containers.intersystems.com/iris-ml-community docker pull containers.intersystems.com/iris-ml-community You can download the IRIS image and samples from https://github.com/KurroLopez/iris-mll-fordummies. 📣TIP: You can open the terminal from Docker with the following command: docker-compose exec iris iris session iris Sleepland University studio Sleepland University has done extensive research on insomnia, conducting thousands of interviews and building a database with various parameters of patients with and without sleeplessness. The collected data includes the following: Gender (male/female) Age (The age of the person in years) Occupation (The occupation or profession of the person) Sleep Duration (The number of hours the person sleeps per day) Quality of Sleep (A subjective rating of the quality of sleep, ranging from 1 to 10) Physical Activity Level (The number of minutes the person engages in physical activity daily) Stress Level (A subjective rating of the stress level experienced by the person, ranging from 1 to 10) BMI Category (The BMI category of the person: Underweight, Normal, Overweight) Systolic (Systolic blood pressure) Diastolic (Diastolic blood pressure) Heart Rate (The resting heart rate of the person in BPM) Daily Steps (The number of steps the person takes per day) Sleep Disorder (None, Insomnia, Sleep Apnea) For the first sample, I created a class (St.MLL.insomniaBase) with the columns mentioned above: Class St.MLL.insonmniaBase Extends %Persistent { /// Gender of patient (male/female) Property Gender As %String; /// The age of the person in years Property Age As %Integer; /// The occupation or profession of the person Property Occupation As %String; /// The number of hours the person sleeps per day Property SleepDuration As %Numeric(SCALE = 2); /// A subjective rating of the quality of sleep, ranging from 1 to 10 Property QualitySleep As %Integer; /// The number of minutes the person engages in physical activity daily Property PhysicalActivityLevel As %Integer; /// A subjective rating of the stress level experienced by the person, ranging from 1 to 10 Property StressLevel As %Integer; /// The BMI category of the person: Underweight, Normal, Overweight Property BMICategory As %String; /// Systolic blood pressure Property Systolic As %Integer; /// Diastolic blood pressure Property Diastolic As %Integer; /// The resting heart rate of the person in BPM Property HeartRate As %Integer; /// The number of steps the person takes per day Property DailySteps As %Integer; /// None, Insomnia, Sleep Apnea Property SleepDisorder As %String; } Then, I built some classes extending from insomniaBase called insomnia01, insomniaValidate01, and insomniaTest01. It allowed me to have the same columns for each table. Eventually, we will need to populate our tables with sample values, so I designed a class method for that purpose. Class St.MLL.insomnia01 Extends St.MLL.insomniaBase { /// Populate values ClassMethod Populate() As %Status { write "Init populate "_$CLASSNAME(),! &sql(TRUNCATE TABLE St_MLL.insomnia01) …… write $CLASSNAME()_" populated",! Return $$$OK } 📣TIP: To open the terminal, type the following command: docker-compose exec iris iris session iris Using the terminal, call the method Populate of this class Do ##class(St.MLL.insomnia01).Populate() If we do everything right, we will have a table with the values for training our ML. We also need to create a new table for validation. It is easy because you will only require a part of the data provided for the training. In this case, it will be 50% of the items. Please, run the following sentence in the terminal. Do ##class(St.MLL.insomniaValidate01).Populate() Finally, we will prepare some test data to see the results of our training. Do ##class(St.MLL.insomniaTest01).Populate() Train, train, and train... you will become stronger Now, we have all the data needed to train our model. How to do it? You will only need 4 simple instructions: CREATE MODEL TRAIN MODEL VALIDATE MODEL SELECT PREDICT Creating the model CREATE MODEL creates the Machine Learning model metadata by specifying the model’s name, the target field to be predicted, and the dataset that will supply the target field. In our sample, we have some parameters to evaluate sleep disorders so we will design the following models: insomnia01SleepModel: By gender, age, sleep duration and quality of sleep. Check if the age and sleeping habits affect any sleep disorder type. insomnia01BMIModel: By gender, age, occupation and BMI category. Examine whether age, occupation and BMI affect any sleep disorder type. insomnia01AllModel: All factors Inspect if all factors affect any sleep disorder type.. We are going to create all those models now. Using SQL management in IRIS portal, type the following sentence: CREATE MODEL insomnia01AllModel PREDICTING (SleepDisorder) From St_MLL.insomnia01 At this point, our model knows which column to predict. You can check what was created and what the predicting column contains with the sentence below: SELECT * FROM INFORMATION_SCHEMA.ML_MODELS Ensure that the predicting column name and the columns themselves are correct. However, we also want to add different model types since we wish to predict sleep disorders according to other factors, not all fields. In this case, we are going to use the "WITH" clause to specify the columns that should be used as parameters to make the prediction. To utilize the "WITH" clause, we must indicate the name of the columns and their type. CREATE MODEL insomnia01SleepModel PREDICTING (SleepDisorder) WITH(Gender varchar, Age integer, SleepDuration numeric, QualitySleep integer) FROM St_MLL.insomnia01 CREATE MODEL insomnia01BMIModel PREDICTING (SleepDisorder) WITH(Gender varchar, Age integer, Occupation varchar, BMICategory varchar) FROM St_MLL.insomnia01 Make sure that all those models have been successfully created. Training the model The TRAIN MODEL command runs the AutoML engine and specifies the data that will be used for training. FROM syntax is generic and allows the same model to be trained multiple times on various data sets. For instance, you may train a table with data from Sleepland University or Napcity University. The most important thing though is to have the data model with the same fields, same name, and the same type. The AutoML engine automatically performs all necessary machine-learning tasks. It identifies relevant candidate features from the selected data, evaluates feasible model types based on the data and problem definition, and sets hyperparameters to create one or more viable models. Since our model has 50 records, it is enough for such training. TRAIN MODEL insomnia01AllModel FROM St_MLL.insomnia01 Do the same with other models. TRAIN MODEL insomnia01SleepModel FROM St_MLL.insomnia01 TRAIN MODEL insomnia01BMIModel FROM St_MLL.insomnia01 You can find out whether your model has been properly trained with the following sentence: SELECT * FROM INFORMATION_SCHEMA.ML_TRAINED_MODELS It is necessary to validate the model and the training with the command VALIDATE MODEL. Validating the model At this stage, we need to confirm that the model has been trained properly. So, we should run the command VALIDATE MODEL. 📣Remember: Before populating the class, validate it with 50% of the data from the training data source. VALIDATE MODEL returns simple metrics for regression, classification, and time series models based on the provided testing set. Check what has been validated with the sentence below: VALIDATE MODEL insomnia01AllModel From St_MLL.insomniaValidate01 Repeat it with other models. VALIDATE MODEL insomnia01SleepModel FROM St_MLL.insomniaValidate01 VALIDATE MODEL insomnia01BMIModel FROM St_MLL.insomniaValidate01 Consuming the model Now, we will consume this model and inspect whether the model has been learning accurately how to produce the Result value. With the help of the sentence “SELECT PREDICT”, we are going to forecast what the value of the Result will be. To do that, we will use the test1 table populated before. SELECT *, PREDICT(insomnia01AllModel) FROM St_MLL.insomniaTest01 The result looks weird after utilizing 50% of the data exploited to train the model... Why has a 29-year-old female nurse been diagnosed with “insomnia”, whereas the model predicted “sleep apnea”? (see ID 54). We should examine other models (insomnia01SleepModel and insomnia01BMIModel), created with different columns, but don't worry! I will display the columns used to design them. SELECT Gender, Age, SleepDuration, QualitySleep, SleepDisorder, PREDICT(insomnia01SleepModel) As SleepDisorderPrediction FROM St_MLL.insomniaTest01 You can see again that a 29-year-old female has been diagnosed with “insomnia”, whereas the prediction states “sleep apnea”. Ok, you are right! We also need to know what percentage of the prediction has been applied to this final value. How can we know the percentage of a prediction? To find out the percentage of the prediction, we should exploit the command “PROBABILITY”. This command retrieves a value between 0 and 1. However, it is not the probability of prediction, it is the probability to get the value that you wish to check. This is a good example: SELECT *, PREDICT(insomnia01AllModel) As SleepDisorderPrediction, PROBABILITY(insomnia01AllModel FOR 'Insomnia') as ProbabilityPrediction FROM St_MLL.insomniaTest01 It is the probability of getting “Insomnia” as a sleep disorder. Our nurse, a woman, 29 years old, diagnosed with “Insomnia” has a 49.71% chance of having Insomnia. Still, the prediction is “Sleep Apnea” … Why? Is the probability the same for other models? SELECT Gender, Age, SleepDuration, QualitySleep, SleepDisorder, PREDICT(insomnia01SleepModel) As SleepDisorderPrediction, PROBABILITY(insomnia01SleepModel FOR 'Insomnia') as ProbabilityInsomnia, PROBABILITY(insomnia01SleepModel FOR 'Sleep Apnea') as ProbabilityApnea FROM St_MLL.insomniaTest01 Finally, it is a bit clearer now. According to the data (sex, age, sleep quality, and sleep duration), the probability of having insomnia is only 34.63%, whereas the chance of having sleep apnea is 64.18%. Wow…It is very interesting! Still, we were exploiting only a small portion of data inserted directly into a table with a class method… How can we upload a huge file with data? Please, wait for the next article!, it is coming soon. Thank you @Francisco.López1549 , it is very deep and easy to read. Have you ever had a real use case to use Machine Learning in your daily work? I just ask due to me curiosity.