The aim of WORK PACKAGE 3 is to assess the quality of ExplainMe technology in specific scenarios. application and implementation in cooperation with a business partner.
(1) Preparing use cases and application scenarios for the ExplainMe system (M1-24, in cooperation with Prof. Leidner and Dr. J. Sousa)
At least 3 application scenarios will be developed for which use cases will be defined in in consultation with specialists in the fields of medicine and law, among others. The first application concerns an important psychiatric problem of monitoring affective disorders based on voice. Use cases will be developed in accordance with Volere methodology [ https://www.volere.org/wpcontent/uploads/2018/12/VolereGettingStarted.pdf, Volere Requirements: How to Get Started], which includes tools to effectively define scenarios business and, among others, identification and analysis of stakeholders and collection of requirements (functional and non-functional) and validation of these requirements (completeness, suitability, testability, consistency, traceability). For each scenario application, the scope of input data will be defined. IBS PAN will provide ExplainMe for the project access to two large, digital, anonymous data sets collected from patients suffering from, among others, bipolar affective disorder, which will be used to conduct experiments. The research was approved Regional Bioethics Committee Medical Chamber in Warsaw (agreement no. KB/1094/17). The result of this task will be specification (its three iterations) D.3.1 . Establishing cooperation with a second economic partner is also expected.
(2) Development of ExplanMoodMon software (M12-M24 in cooperation with a business partner) Britenet Med sp z o. o.)
The task will involve developing an ExplainMoodMon solution for Britenet Med. As part of previous work and including the project POIR.01.01.01-00-0342/20-00 “Development of a system supporting monitoring of the course and early detection of affective disorder relapses based on artificial intelligence algorithms” the company created a system MoodMon used to monitor patients with affective disorders. Unfortunately, low trust remains a problem among users due to the lack of explanations for alarms. The result of this task will be the ExplainMoodMon system D.3.2. fully integrated with MoodMon software.
(3) Validation of ExplainMe technology in application scenarios (M18-48)
Development of a test and validation plan for real-world data in application scenarios as needed an updated reference document defining the process for testing and evaluating the ExplainMe system and its modules.
The plan includes the identification of KPIs (with the participation of end users) and ethical, social, related to gender and legal. The output of this task will be D.3.4. collecting validation results for application scenarios in this implementation.