ABSTRACT
Machine learning models can use information from gene expressions in patients to efficiently predict the severity of symptoms for several diseases. Medical experts, however, still need to understand the reasoning behind the predictions before trusting them. In their day-to-day practice, physicians prefer using gene expression profiles, consisting of a discretized subset of all data from gene expressions: in these profiles, genes are typically reported as either over-expressed or under-expressed, using discretization thresholds computed on data from a healthy control group. A discretized profile allows medical experts to quickly categorize patients at a glance. Building on previous works related to the automatic discretization of patient profiles, we present a novel approach that frames the problem as a multi-objective optimization task: on the one hand, after discretization, the medical expert would prefer to have as few different profiles as possible, to be able to classify patients in an intuitive way;on the other hand, the loss of information has to be minimized. Loss of information can be estimated using the performance of a classifier trained on the discretized gene expression levels. We apply one common state-of-the-art evolutionary multi-objective algorithm, NSGA-II, to the discretization of a dataset of COVID-19 patients that developed either mild or severe symptoms. The results show not only that the solutions found by the approach dominate traditional discretization based on statistical analysis and are more generally valid than those obtained through single-objective optimization, but that the candidate Pareto-optimal solutions preserve the sense-making that practitioners find necessary to trust the results. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
ABSTRACT
This article presents the analysis and design of the intelligent agent model “IA-ACR”, which has the objective of monitoring movements, which are carried out in a coordinated and intelligent way in a robot, which will have the task of performing routines of physical exercises and dance, these routines will then be imitated by children with neurodevelopmental disorders (NDD), in order to capture their attention so that therapies are more effective, which will be evaluated by the specialist (psychologist). Due to the current situation of the pandemic that is being experienced due to COVID-19, health protocols were established, such as avoiding contact between people, given this restriction, a digital platform was developed that serves as support for children in order to receive their sessions, where the robot appears through videos, this being an advantage of telehealth. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.