@article{ART003348225},
author={Kim young in},
title={Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection},
journal={ Journal of Software Forensics},
issn={3092-541X},
year={2026},
volume={22},
number={2},
pages={129-142},
doi={10.29056/jsf.2026.06.12}
TY - JOUR
AU - Kim young in
TI - Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection
JO - Journal of Software Forensics
PY - 2026
VL - 22
IS - 2
PB - Korea Software Assessment and Valuation Society
SP - 129
EP - 142
SN - 3092-541X
AB - Handwriting analysis is a non-invasive method for early detection of Alzheimer's disease. Prior DARWIN dataset studies reported only simple metrics such as accuracy and AUC, without addressing the reliability of predicted probabilities or the handling of uncertain cases. This study re-evaluates six machine learning models using repeated 5×5 cross-validation, applying four probability calibration techniques and Mondrian Inductive Conformal Prediction to quantify uncertainty. Random Forest showed the highest performance (AUC 0.957) but the largest calibration error (ECE 0.192). Isotonic regression reduced this by 53% (ECE 0.090), a significant improvement (p<0.001), and a rejection option withheld predictions for uncertain cases. Decision curve analysis showed that calibrated probabilities can yield greater net benefit for clinical decision-making. Thus, clinical use of such models requires probability calibration and uncertainty quantification, beyond simple metrics.
KW - Alzheimer’s disease;Handwriting analysis;DARWIN;Probability calibration;Conformal prediction;Decision curve analysis
DO - 10.29056/jsf.2026.06.12
ER -
Kim young in. (2026). Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection. Journal of Software Forensics, 22(2), 129-142.
Kim young in. 2026, "Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection", Journal of Software Forensics, vol.22, no.2 pp.129-142. Available from: doi:10.29056/jsf.2026.06.12
Kim young in "Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection" Journal of Software Forensics 22.2 pp.129-142 (2026) : 129.
Kim young in. Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection. 2026; 22(2), 129-142. Available from: doi:10.29056/jsf.2026.06.12
Kim young in. "Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection" Journal of Software Forensics 22, no.2 (2026) : 129-142.doi: 10.29056/jsf.2026.06.12
Kim young in. Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection. Journal of Software Forensics, 22(2), 129-142. doi: 10.29056/jsf.2026.06.12
Kim young in. Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection. Journal of Software Forensics. 2026; 22(2) 129-142. doi: 10.29056/jsf.2026.06.12
Kim young in. Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection. 2026; 22(2), 129-142. Available from: doi:10.29056/jsf.2026.06.12
Kim young in. "Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection" Journal of Software Forensics 22, no.2 (2026) : 129-142.doi: 10.29056/jsf.2026.06.12
