How Artificial Intelligence Can Be Used to Develop Clinical Outcome Assessments?

PURPOSE: There is a need for the development of valid and reliable clinical outcome assessments to support earlier disease detection and better evaluation of patients’ symptoms, experience, and burden. With the support of emerging methods using Artificial Intelligence (AI), we are able to analyze large amounts of patient generated health data in order to develop these clinical outcome assessments. In this session, we will explore how Dr. Pirge, Dr. Kwon and Dr. Mckay are using AI in their field of research to develop predictive algorithms and reliable assessments of clinical outcomes and how it could improve precision of clinical trials and access to care.

DESCRIPTION:

Lysel Brignoli will open the workshop by talking about the need for the development of valid and reliable clinical outcome assessments and the issues the industry is facing when developing these assessments.

Dr. Pirge will discuss how he accesses clinical notes via Natural Language Processing (NLP) and Large Language Models (LLM) to assess clinical outcomes. He will address how to best combine NLP and LLM models. He will describe examples that John Snow Labs has delivered in the areas of mental health and Oncology.

Dr. Kwon and Dr. Mckay will describe an example that applied state-of-the-art deep learning approaches to score freezing of gait (FOG) in people with Parkinson’s disease (PD). Currently, FOG is a poorly understood heterogeneous gait disorder which is measured subjectively by movement disorders specialists or through patient completed questionnaires that are unsuitable for use in clinical trials.

The discussants will facilitate a discussion with the audience about their specific needs for patient centered clinical outcome assessments, and also the expertise and resources needed to use AI in this field and the potential use in the future.