ICON-D: Assessing Chess as a Predictor of Neurodegenerative Risk
Nithin Ramasamy·Psychology·July 13, 2026·WY-2026-0002
The ICON-D study explores whether chess activity and history can serve as neurocognitive markers for neurodegenerative disease. In the present aging society, where birth rates are declining, and 91% of neurological disorder cases affect the elderly, the population faces an increasing risk. Early diagnosis is crucial but difficult because of the subtle symptoms. If chess engagement can serve as a predictor, it may support pre-diagnosis and enable earlier intervention. 52 participants completed chess puzzles and cognitive tests like the Tower of London, Working Memory, and Wisconsin Card Sorting, which assess executive function, planning, and problem-solving. Data collected included puzzle time, accuracy, weekly chess hours (WCH), and diagnosis status. Machine learning models such as Random Forest, XGBoost, and SVM are used to analyze how these features relate to neurological risk. Feature importance and clustering identified key predictors. ROC-AUC scores evaluated model performance. Random Forest achieved the highest ROC-AUC (.9947), followed by XGBoost (.9891) and SVM (.9824). Age and WCH consistently ranked as the top predictors. Random Forest showed Age (53.58%) and WCH (22.10%) as most influential; XGBoost ranked Age (49.82%) and WCH (41.65%) highest. Weekly chess activity proved key across all models. The ICON-D study shows that machine learning can use chess activity, especially weekly playtime, along with age and cognitive data to predict neurodegenerative risk. With high model accuracy, chess offers a cost-effective, non-invasive tool for early diagnosis and intervention