The goal of this clinical trial is to validate an artificial intelligence algorithm called CAUD-PIN (CAUchemar Detection on INdividualized Parameters) to predict nightmare-related awakenings in military personnel and veterans suffering from post-traumatic stress disorder (PTSD). The main question it aim to answer is :
* Can the CAUD-PIN algorithm detect physiological patterns preceding trauma-related nightmare awakenings with a sensitivity greater than 80%? * What physiological signals best predict the occurrence of a trauma-related nightmare awakening?
Participants will:
Wear two non-invasive devices at home - an EEG headband and a wrist-worn sensor bracelet - for 15 consecutive nights Press a button on the bracelet each time they wake up from a nightmare Complete sleep and daytime diaries every evening and morning Attend two clinical assessments (at enrollment and at the end of the study, one month later) at one of the three participating military hospitals
The algorithm will be applied to the recorded data after collection and does not intervene in real time or affect participants' standard medical care in any way.
Age between 18 and 65 years (age limit related to military status)
Active military personnel or veterans
Receiving psychiatric care in a military hospital (HNIA Bégin, HNIA Percy, or HRIA Brest) for PTSD related to combat trauma (exposure to death or life-threatening situations)
PCL-5 score > 33
Pre-existing progressive psychiatric disorder prior to trauma (condition requiring long-term treatment > 1 year)
Sleep disorder affecting sleep architecture (obstructive sleep apnea syndrome, narcolepsy)
Neurological disorder (acute confusional state, severe traumatic brain injury) Inability to provide informed consent
Persons under legal protection measures (guardianship, curatorship, or equivalent judicial protection)
Single group
CAUD-PIN (CAUchemar Detection on INdividualized Parameters) - AI-based predictive algorithm forTrauma-related nightmare awakenings
Diagnostic testCAUD-PIN (CAUchemar Detection on INdividualized Parameters) is a non-invasive, AI-based predictive algorithm developed by the VIFASOM research team (UMR 7330, Université Paris Cité / French Military Health Service) from the SOMMEPT clinical project. It analyzes multimodal physiological signals recorded during home sleep to identify patterns preceding Trauma-related nightmare awakenings in military patients with PTSD. Data are collected using two wearable devices: an EEG headband (Hypnodyne Zmax Lite) measuring brain activity and sleep stages, and a wrist-worn multisensor bracelet (Empatica EmbracePlus) measuring heart rate, electrodermal activity, skin temperature, and motor activity. The algorithm is applied exclusively post-hoc to pseudonymized recordings, with no real-time intervention and no impact on participants' standard medical care.
Sensitivity of the CAUD-PIN algorithm in predicting Trauma-related nightmare awakenings
Sensitivity will be calculated as the number of Trauma-related nightmare awakenings correctly identified by the CAUD-PIN algorithm divided by the total number of confirmed trauma-related nightmare awakenings, multiplied by 100. Unit of Measure: Percentage
Specificity of the CAUD-PIN algorithm for predicting Trauma-related nightmare awakenings
Specificity will be calculated as the number of events correctly classified as not being trauma-related nightmare awakenings divided by the total number of confirmed non-nightmare events, multiplied by 100. Unit of Measure: Percentage
Positive predictive value of the CAUD-PIN algorithm
Positive predictive value will be calculated as the number of correctly predicted trauma-related nightmare awakenings divided by the total number of trauma-related nightmare awakenings predicted by the CAUD-PIN algorithm, multiplied by 100. Unit of Measure: Percentage
Negative predictive value of the CAUD-PIN algorithm
Negative predictive value will be calculated as the number of events correctly predicted as not being trauma-related nightmare awakenings divided by the total number of events predicted as not being trauma-related nightmare awakenings, multiplied by 100. Unit of Measure: Percentage
Prediction lead time before trauma-related nightmare awakening
Mean time (in seconds) between the physiological predictive pattern detected by the CAUD-PIN algorithm and the participant's button press indicating a trauma-related nightmare awakening. Unit of Measure: Seconds
Click on the lead sponsor to view all of their trials.
