Back

A Blood-Based Transcriptomic Signature for PTSD Classification Using Machine Learning

Alipour, S.; Pamanji, R.; Jamil, E.; Yeguvapalli, S.; Chitrala, K. N.

2026-05-13 health informatics
10.64898/2026.05.10.26352854 medRxiv
Show abstract

Post-traumatic stress disorder (PTSD) remains a significant psychiatric burden; despite growing biomarker research, no blood-based molecular diagnostic tool has been clinically validated for routine use. In this study, we developed a machine learning classifier for PTSD using peripheral blood leukocyte RNA-seq data from combat-exposed U.S. Marines (GSE64813), diagnosed via the Clinician-Administered PTSD Scale (CAPS) under DSM-IV criteria. Differentially expressed genes (DEGs) were identified and further refined through additional filtering criteria, yielding a 90-gene feature set used to train and compare multiple machine learning models. The support vector machine (SVM) classifier achieved the best performance, with an accuracy of 89% and an AUC of 0.95, outperforming logistic regression and random forest approaches. Furthermore we evaluated our model on independent external datasets to assess generalizability. These findings highlight the promise of transcriptomic signatures as a foundation for objective, blood-based PTSD diagnostics, while emphasizing the critical need for robust cross-dataset generalizability. Code availabilityhttps://www.kaggle.com/code/persianexxx/ptsd-final

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.