Data-Driven Modelling of Subjective Pain/Pleasure Assessments As Responses to Vaginal Dilation Stimuli

D. Varagnolo*, S. Knorn, E. Oliver-Chiva, R. Melles, M. Dewitte

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Women affected by pain during penetrative sexual intercourse are often treated using fixed-size vaginal dilators that are regularly perceived as uncomfortable and leading to premature treatment drop-outs. These dilators could be improved by making them adaptive, i.e., able to exert dynamically different pressures on the vaginal duct to simultaneously guarantee comfort levels and achieve the medical dilation objectives. Implementing feedback control would then benefit from models that connect the patients' comfort levels with their experienced physiological stimuli. Here we address the problem of data-driven quantitative modeling of pain/pleasure self-assessments obtained through medical trials. More precisely, we consider time-series records of pelvic floor muscles pressure, vaginal dilation, and pain/pleasure evaluations, and model the relations among these quantities using statistical analysis tools. Besides this, we also address the important issue of the individualization of these models: different persons may respond differently, but these variations may sometimes be so small that it may be beneficial to learn from several individuals simultaneously. We here numerically validate the previous claim by verifying that clustering patients in groups may lead, from a data-driven point of view, to models with a significantly improved statistical performance.
Original languageEnglish
Pages (from-to)423-428
Number of pages6
JournalIEEE Control Systems Letters
Volume2
Issue number3
DOIs
Publication statusPublished - 1 Jul 2018

Keywords

  • Dyspareunia
  • modelling of psychological systems
  • support vector classification
  • clustering

Cite this