Hyperparameter-tuned prediction of somatic symptom disorder using functional near-infrared spectroscopy-based dynamic functional connectivity
Objective. Somatic symptom disorder (SSD) is a reflection of medically unexplained physical symptoms that lead to distress and impairment in social and occupational functioning. SSD is phenomenologically diagnosed and its neurobiology remains unsolved. Approach. In this study, we performed hyper-par...
| Autores: | , , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2019 |
| País: | España |
| Institución: | Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya) |
| Repositorio: | Recercat. Dipósit de la Recerca de Catalunya |
| OAI Identifier: | oai:recercat.cat:10230/43925 |
| Acceso en línea: | http://hdl.handle.net/10230/43925 http://dx.doi.org/10.1088/1741-2552/ab50b2 |
| Access Level: | acceso abierto |
| Palabra clave: | Dynamic functional connectivity fNIRS Hyperparameter optimization Machine learning Somatic symptom disorder |
| Sumario: | Objective. Somatic symptom disorder (SSD) is a reflection of medically unexplained physical symptoms that lead to distress and impairment in social and occupational functioning. SSD is phenomenologically diagnosed and its neurobiology remains unsolved. Approach. In this study, we performed hyper-parameter optimized classification to distinguish 19 persistent SSD patients and 21 healthy controls by utilizing functional near-infrared spectroscopy via performing two painful stimulation experiments, individual pain threshold (IND) and constant sub-threshold (SUB) that include conditions with different levels of pain (INDc and SUBc) and brush stimulation. We estimated a dynamic functional connectivity time series by using sliding window correlation method and extracted features from these time series for these conditions and different cortical regions. Main results. Our results showed that we found highest specificity (85%) with highest accuracy (82%) and 81% sensitivity using an SVM classifier by utilizing connections between right superior temporal–left angular gyri, right middle frontal (MFG)—left supramarginal gyri and right middle temporal—left middle frontal gyri from the INDc condition. Significance. Our results suggest that fNIRS may distinguish subjects with SSD from healthy controls by applying pain in levels of individual pain-threshold and bilateral MFG, left inferior parietal and right temporal gyrus might be robust biomarkers to be considered for SSD neurobiology. |
|---|