Next-Gen Agent App Powered by Explainable AI

This project is a system for identifying dangerous diseases, including COVID-19, based on a combination of features: temperature, blood pressure, and external signs of the disease.

Comment

Inspiration

It is necessary to promptly identify epidemics of dangerous diseases. This allows for their spread to be quickly localized. Another problem is monitoring patients requiring constant care using known trackers. This makes it possible to provide them with timely medical care. Medicine requires unambiguous and reliable identification of diseases based on objective parameters of sensors and trackers, which excludes the use of so-called artificial intelligence based on LLM, which sometimes generates false information based on objective data.

What it does

Using objective parameters measured by trackers, sensors or entered by the user, it processes them without using logical abstract reasoning and gives the most probable disease from the diseases entered into our database based on directly measured parameters and direct observations, which are official signs of diseases listed in open medical databases for these diseases, such as MSD Manuals (msdmanuals.com) and others. Therefore, the obtained result at the output will correspond to the conclusion obtained using the provided databases and reference books with a fully explainable solution that the attending physicians could come to, limited by the information provided in the said databases.

How we built it

The application was built on the basis of a working model of explainable artificial intelligence, first published in the work of Dubovikov N.M. Mathematical model of social - innovative economy: articles / Dubovikov N.M. - LAP LAMBERT Academic Publishing, 2013. - P. 114: Mathematical formalization of the procedure for obtaining and accumulating new information through analysis and synthesis, certificate of authorship UA № 91811, which does not require preliminary training or requires it in a minimal amount with a predictable outcome of this training. In particular, with the help of this model, the world's only working application for identifying industrial designs at the MVP stage was created, which has no competitors in the world, which practically does not require preliminary training. The proposed application Next-Gen Agent App Powered by Explainable AI is written in Python and is located in the author's repository on GitHub https://github.com/Mykola2056/Next-Gen-Agentic-App/blob/main/Next-GenAgentAppXAI%20Python310.py. Explanations of the program's operation are on the disk at the link https://drive.google.com/file/d/1JwQk8AZdxGze43xg4809Uf6RY8izBcys/view And on the video at https://youtu.be/Urp6h6LHGck Below is a picture Jupyter Notebook: Jupyter Notebook Next-Gen Agent App Powered by Explainable AI Python 3.10 https://drive.google.com/file/d/1LSmYt4qTWgBqm2oKmy5VJe_lvHbUesx4/view?usp=drive_link

Challenges we ran into

The main problem for writing this application was the lack of the necessary specially structured databases of objective measurable parameters that would allow their direct use for this application.

Accomplishments that we're proud of

A working application was created based on the principles of explainable artificial intelligence, confirming the operability of the mathematical model described in the work of Dubovikov N.M. Mathematical model of social and innovative economy: articles / Dubovikov N.M. - LAP LAMBERT Academic Publishing, 2013. - P. 114: Mathematical formalization of the procedure for obtaining and accumulating new information through analysis and synthesis. The application can be easily integrated into any structured database of disease signs with any number of objective parameters recorded by trackers and sensors providing data for identifying the corresponding disease.

What we learned

The application is created taking into account the information in the ArangoDB and other medical professional databases and the data in the application is presented in a graphical format. The simplicity of the application is such that it does not require the use of NVIDIA technology, which is its advantage compared to L LM models. Accepts requests in natural language. Clearly and intuitively visualizes the results, returning them to the interface in natural language.

What's next for Next-Gen Agent App Powered by Explainable AI

We can implement projects of special structuring of medical databases for their processing by our explainable artificial intelligence, integration with existing official databases of disease signs and methods of their treatment in accordance with the International Classification of Diseases ICD-10, the International Classification of Disease Symptoms, classification of disease signs and a set of data parameters. Our application can be expanded to any number of diseases and the corresponding number of signs of these diseases, determined using appropriate medical measuring instruments collecting information through various channels and sensors, including: optical, infrared, X-ray and other information in a systematic manner.

Built With

Share this project:

Updates