ML (Machine Learning) ethical and legal issues in automatic categorization of FHIR Patient profiles through SMART on FHIR agent
BRIEF SUMMARY OF THE PRESENTATION
The automatic diagnosis of abnormal health conditions, resulting from unusual health impacting events can be developed through a machine learning approach. Equiped with synthetic artificial datasets that represent realistic features of signals, symptoms and other environmental factors that can be of interest in the diagnosis in 3 areas considered: COPD, Diabetes and Cardiovascular System Disease, a highly accurate classifier was developed by the team. The synthetic datasets are full FHIR representations that were targeted to real population distributions of the above mentioned health problems that are very significative in the current occidental epidemic reality. Automated Machine Learning techniques were used to achieve state of the art accuracy, precision and recall that are presented and discussed in this article. Our work is part of the wider effort to develop the ICP (Individual Care Process) where healthcare alerts are automatically issued to the less diferentiated caregiver for providing the proper adequate care through continuous health monitoring.
The development options like dataset definition, feature reduction, AutoML tooling election and execution are presented, evaluated and results discussed.
A sound scientific approach was developed to demostrate that the resulting system outperforms any clinical professional equiped with the same data both in diagnosys quality and speed.
The present talk elaborates about the ethical and legal issues that arise when aplying AI techniques, mainly AML (Automated Machine Learning), to realistic synthetic datasets that represent any real population distribution.
- Health representation standards and FHIR concretely
- Smart on FHIR as an innovatove tool for this decade
- Creation of realistic synthetic datasets for a population
- Profiling in the dataset using AutoML
- Ethical and legal issues
David Mendes
BRIEF RESUME OF THE SPEAKER
David Mendes currently works at Universidade de Évora and at the Politécnico de Santarém.
David does research in Artificial Intelligence and Computing in Mathematics, Natural Science, Engineering and Medicine. His current project is 4IE – Instituto Internacional de Investigação e Inovação do Envelhecimento’.