Voice recordings taken prior to the start of repetitive transcranial magnetic stimulation (rTMS) therapy will be used to develop a predictive model of treatment response using machine learning for patients with major depressive disorders (MDD).
The objective of this study is to investigate whether voice biomarkers taken prior to the start of rTMS can predict the response to a 30-session course of rTMS in patients with MDD with good sensitivity and specificity. Patients who are prescribed rTMS to treat an episode of MDD at one of the nine participating centres will be invited to participate in the study. The protocol includes one inclusion visit and seven evaluation visits before, during, and after the course of rTMS. The rTMS course will last for six weeks, comprising of 30 sessions with five sessions per week. Prior to the commencement of the rTMS course, patients will need to complete several questionnaires. Before the first session, voice tests will be recorded using the Callyope application. Patients will answer questionnaires and assess voice tests weekly and after the final rTMS session to track the progression of depressive symptoms.
Study Type
OBSERVATIONAL
Enrollment
156
rTMS will be applied 5 days out of 7 for 6 weeks, for a total of 30 sessions. The area stimulated will be the left dorsolateral prefrontal cortex, at a frequency of 10 Hz, at 110% of motor threshold for approximately 20 minutes (i.e. 3,000 stimulations per session).
Clinique Inicea du Pays de Seine
Bois-le-Roi, France
Clinique Inicea La Mare O Dans
Les Damps, France
Clinique Inicea Villa des Roses
Lyon, France
Etablissement public de santé de Ville-Evrard
Area under the curve of the model predicting response to rTMS treatment
Area under the curve of the model predicting response to rTMS treatment, objectified by a 50% reduction in MADRS depression score, trained from vocal biomarkers recorded before rTMS treatment.
Time frame: 6 weeks : before the first session of rTMS (vocal samples) and before and after the rTMS treatment (MADRS score)
Area under the curve of the rTMS remission prediction model
Area under the curve of the rTMS remission prediction model, objectified by a post-course MADRS score lower than 10, trained from vocal biomarkers recorded before the rTMS course.
Time frame: 6 weeks : before the first session of rTMS (vocal samples) and before and after the rTMS treatment (MADRS score)
Area under the curve of the prediction model of the self-assessed response to the rTMS course
Area under the curve of the prediction model of the self-assessed response to the rTMS course, objectified by a 50% reduction in PHQ-9 depression score, trained from the vocal biomarkers recorded before the rTMS course.
Time frame: 6 weeks : before the first session of rTMS (vocal samples) and each week during the rTMS treatment (PHQ-9 score)
Area under the curve of the prediction model of the response to the rTMS course, trained with the addition of weekly recorded voice samples.
Area under the curve of the prediction model of the response to the rTMS course, trained with the addition in the initial model of vocal biomarkers recorded each week of the rTMS course.
Time frame: 6 weeks : weekly
Area under the curve of the model predicting the response to rTMS treatment, trained with the addition of clinical indicators
Area under the curve of the model predicting the response to rTMS treatment, trained with the addition in the initial model of clinical indicators other than depression.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.
Neuilly-sur-Marne, France
Centre médical de psychiatrie NeuroStim Luxembourg
Paris, France
Centre médical de psychiatrie NeuroStim Etoile
Paris, France
Clinique Inicea Jeanne d'Arc Hopital Privé Parisien
Saint-Mandé, France
Centre montois de psychiatrie ambulatoire Inicea
Saint-Pierre-du-Mont, France
Centre Tourangeau de psychiatrie ambulatoire Inicea
Tours, France
Time frame: 6 weeks : before the first session of rTMS (vocal samples) and before and after or weekly (clinical indicators)
Comparison between the scores predicted by the model and the actual scores obtained on the different questionnaires assessed.
Coefficients of determination R² and average absolute errors between the scores predicted by the model and the actual scores obtained on the different PHQ-9, GAD-7, IADL, AIS and MAP-SR questionnaires.
Time frame: 6 weeks : before the first session of rTMS (vocal samples) and before and after the treatment (questionnaires)
Area under the curve of the different prediction models obtained, starting from pre-trained models on healthy subject databases.
Area under the curve of the different prediction models obtained, starting from pre-trained models on healthy subject databases.
Time frame: 6 weeks : Before and after rTMS treatment
Risk and coverage of the different prediction models obtained.
Risk and coverage of the different prediction models obtained.
Time frame: 6 weeks : After rTMS treatment
Comparison of the prediction models obtained with or without the use of advanced data encryption algorithms.
Comparison of the areas under the curve of the different prediction models obtained with or without the use of advanced data encryption algorithms.
Time frame: 6 weeks : After rTMS treatment