Keywords: IRIS, Agents, Agentic AI, Smart Apps
Transformer based LLMs appear to be a pretty good "universal logical–symbolic abstractor". They started to bridge up the previous abyss among human languages and machine languages, which in essence are all logic symbols that could be mapped into the same vector space.
Wondering for 3 years we might be able to just use English (etc human natural languages) to do IRIS implementations as well, one day.
Keywords: Vibe coding, Windsurf, IRIS, TIE
Has anyone not been trying "vibe coding" so far?
Even merely 3 years ago, if anyone asked
Keywords: ChatGPT, COS, Lookup Table, IRIS, AI
Here is another quick note before we move on to GPT-4 assisted automation journey. Below are some "little" helps ChatGPT had already been offering, here and there, during daily works.
And what could be the perceived gaps, risks and traps to LLMs assisted automation, if you happen to explore this path too. I'd also love to hear anyone's use cases and experiences on this front too.
One of the simplest tasks could be Lookup tables.
I tried OpenAI GPT's coding model a couple of weeks ago, to see whether it can do e.g. some message transformations between healthcare protocols. It surely "can", to a seemingly fair degree.
It has been nearly 3 weeks, and it's a long, long time for ChatGPT, so I am wondering how quickly it grows up by now, and whether it could do some of integration engineer jobs for us, e.g. can it create an InterSystems COS DTL tool to turn the HL7 into FHIR message?
Immediately I got some quick answers, in less than one minute or two.
Large language models are stirring up some phenomena in recent months. So inevitably I was playing ChatGPT too over last weekend, to probe whether it would be a complimentary to some BERT based "traditional" AI chatbots I was knocking up, or rather would it simply sweep them away.
A thought comes to mind while playing. By going slightly theoretical or philosophical, eventually interoperability standards such as HL7 and FHIR etc are kind of "languages", right? HL7 has its own grammar, rules, vocabulary and even dialects - every system speaks its own tone.
Keywords: IRIS, IntegratedML, Flask, FastAPI, Tensorflow Serving, HAProxy, Docker, Covid-19
We touched on some quick demos of deep learning and machine learning over the past few months, including a simple Covid-19 X-Ray image classifier and a Covid-19 lab result classifier for possible ICU admissions. We also touched on an IntegratedML demo implementation of the ICU classifier.
Keywords: IRIS, IntegratedML, Machine Learning, Covid-19, Kaggle
Continued from the previous Part I ... In part I, we walked through traditional ML approaches on this Covid-19 dataset on Kaggle.
In this Part II, let's run the same data & task, in its simplest possible form, through IRIS integratedML which is a nice & sleek SQL interface for backend AutoML options. It uses the same environment.
integredML-demo-template defined various ways to load data into
Keywords: IRIS, IntegratedML, Machine Learning, Covid-19, Kaggle
Recently I noticed a Kaggle dataset for the prediction of whether a Covid-19 patient will be admitted to ICU. It is a spreadsheet of 1925 encounter records of 231 columns of vital signs and observations, with the last column of "ICU" being 1 for Yes or 0 for No. The task is to predict whether a patient will be admitted to ICU based on known data.
This dataset seems to be a good example of what's called "traditional ML" task. The data seem to have the right quantity and relatively right quality.
Keyword: Pandas DataFrame, IRIS, Python, JDBC
Pandas DataFrame is popular tool for EDA (Exploratory Data Analysis). In ML tasks, the first thing we usually perform is to understand the data a bit more. Last week I was trying this Covid19 dataset in Kaggle Basically the data is a spreadsheet of 1925 encounter rows with 231 columns, and the task is simply to predict whether a patient (linked to 1 or more encounter records) would be admitted to ICU. So it's a normal classification task, and we would as usual use padas.DataFrame to take a quick look first.
Keywords: PyODBC, unixODBC, IRIS, IntegratedML, Jupyter Notebook, Python 3
A few months ago I touched on a brief note on "Python JDBC connection into IRIS", and since then I referred to it more frequently than my own scratchpad hidden deep in my PC. Hence, here comes up another 5-minute note on how to make "Python ODBC connection into IRIS".
ODBC and PyODBC seem pretty easy to set up in a Windows client, yet every time I stumbled a bit somewhere on setting up an unixODBC and PyODBC client in a Linux/Unix-style server.
Keywords: Deep Learning, Grad-CAM, X-Ray, Covid-19, HealthShare, IRIS
Over the Easter Weekend I touched on some deep learning classifier for Covid-19 Lungs. The demo result seems fine, seemingly matching some academic research publications around that time on this topic. But is it really "fine "?
Recently I happened to listen to an online lunch webinar on "Explainability in Machine Learning", and Don talked about this classification result at the end of his talk:
The above figure is also presented in this research paper: “Why Should I Trust You?
Keywords: COVID-19, Medical Imaging, Deep Learning, PACS Viewer, and HealthShare.
We are all gripped by this unprecedented Covid-19 pandemic. While supporting our customers in battlefields by any means, we also observed various fighting fronts against Covid-19 by leveraging today's AI powers.
Last year I briefly touched a deep learning demo environment.
Keywords: Python, JDBC, SQL, IRIS, Jupyter Notebook, Pandas, Numpy, and Machine Learning
This is another 5-minute simple note on invoking the IRIS JDBC driver via Python 3 within i.e. a Jupyter Notebook, to read from and write data into an IRIS database instance via SQL syntax, for demo purpose.
Last year I touched on a brief note on Python binding into a Cache database (section 4.7) instance.
This is a 10-minute simple step-by-step guide on how to quickly set up various flavors of HealthShare docker containers from scratch on a Win10 laptop.
For example, we can build a couple of HealthShare "global edition vs UK Edition" demos as shown below.
There are a couple of frequently asked questions from HealthShare colleagues and partners:
Keywords: Jupyter Notebook, Tensorflow GPU, Keras, Deep Learning, MLP, and HealthShare
In previous"Part I" we have set up a deep learning demo environment. In this "Part II" we will test what we could do with it.
Many people at my age had started with the classic MLP (Multi-Layer Perceptron) model. It is intuitive hence conceptually easier to start with.
So let's try a Keras "deep learning MLP" with standard demo data that everybody in AI/NN community has been using. It is a kind of so called "supervised learning".
Keywords: Anaconda, Jupyter Notebook, Tensorflow GPU, Deep Learning, Python 3 and HealthShare
This "Part I" is a quick record on how to set up a "simple" but popular deep learning demo environment step-by-step with a Python 3 binding to a HealthShare 2017.2.1 instance . I used a Win10 laptop at hand, but the approach works the same on MacOS and Linux.
Last week it was noticed that Python overtook Java by becoming the most popular language in PYPL Index Tensorflow is a powerful computation engine, very popular in research and academic worlds too.
Recently we supported a few NHS cases that required TIE (Trust Integration Engine) integration with the PKB service. Hence this article is meant to be a 10-minute quick guide to describe a demo solution (simple configurations and end-2-end implementation steps) for Health Connect (Ensemble) Integration with PKB (Patient-Knows-Best) service.