Machine learning-based analysis predicts a person's gender based on mechanical, but not thermal, pain thresholds

Abstract

Abstract Gender differences in pain perception have been extensively studied, while precision medicine applications such as gender-specific pain pharmacology have barely progressed beyond proof-of-concept. A data set comprising pain thresholds to mechanical (blunt and punctate pressure) and thermal (heat and cold) stimuli applied to nonsensitized and sensitized (capsaicin, menthol) forearm skin of 69 male and 56 female healthy volunteers was analyzed for data structures contingent with the prior gender structure, using unsupervised and supervised approaches. A working hypothesis that the relevance of gender differences could be approached via reversibility of the association, i.e., genders should be identifiable from pain thresholds, was verified with trained machine-learning algorithms that could infer a person’s gender in a 20% validation sample not seen to the algorithms during training, with a balanced accuracy of up to 79%. This was only possible with thresholds for mechanical stimuli, but not for thermal stimuli or responses to sensitization, which were not sufficient to train an algorithm that could assign gender better than by guessing or when trained with nonsense (permuted) information. This enabled translation to the molecular level of nociceptive targets that convert mechanical but not thermal information into signals that are interpreted as pain, which could eventually be used for pharmacological precision medicine approaches to pain. By exploiting a key feature of machine learning that enables the recognition of data structures and the reduction of information to the bare minimum relevant, experimental human pain data could be characterized in a way that incorporates "non" logic that could be transferred directly to the molecular pharmacological level, pointing a way toward gender-specific precision medicine for pain.