Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 58 min ago
Peer-reviewedCC BY 4.0Journal of Medical Internet Research

Machine Learning Classification of Bipolar Disorder Across Mood States Using Natural Language Processing and Speech Analysis: Observational Study

  • Yusei Yamaguchi1
  • Yoshinobu Kano2
  • Genichi Sugihara1
  • Ryo Takemura3
  • Hironobu Nakamura1
  • Masaaki Shimizu1
  • Mari Iizuka4
  • Momoko Kitazawa5
  • Ayako Sento5
  • Hidehiko Takahashi1,6
  • Taishiro Kishimoto5
6 affiliations
  1. 1Department of Psychiatry and Behavioral Sciences, Graduate School of Medical and Dental Sciences, Institute of Science Tokyo, 1-5-45 Yushima, Bunkyo-ku, Tokyo, Japan
  2. 2Faculty of Informatics, Shizuoka University, Chuo-ku, Hamamatsu, Shizuoka, Japan
  3. 3Clinical and Translational Research Center, Keio University Hospital, Shinjuku-ku, Tokyo, Japan
  4. 4Department of Neuropsychiatry, Keio University School of Medicine, Shinjuku-ku, Tokyo, Japan
  5. 5Center for the Promotion of Interdisciplinary Research in Medicine and Life Science, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo, 160-8582, Japan, 81 3-5360-1524
  6. 6Center for Brain Integration Research, Institute of Science Tokyo, Tokyo, Japan
Read full textDOICite

Abstract

Background: Bipolar disorder (BP) affects 1% to 3% of adults worldwide, with diagnostic delays averaging 6 years and high initial misdiagnosis rates. The absence of objective diagnostic measures contributes to clinical uncertainty and suboptimal outcomes.

Objective: This study aimed to (1) develop a machine learning model for BP diagnosis and mood state classification using linguistic and acoustic features, (2) identify features associated with different mood states by comparing each with healthy controls (HC), and (3) explore whether features distinguish between mood states within BP.

Methods: We analyzed data prospectively collected over 7 years from 61 patients with BP and 99 HC across Japan. Speech samples were collected via semistructured interviews over 1 to 5 visits and analyzed cross-sectionally, using aggregated or single-visit observations per participant rather than within-person longitudinal modeling. We extracted 6373 acoustic features (openSMILE 3.0, ComParE 2016) and 211 linguistic features. Least absolute shrinkage and selection operator (LASSO)–regularized logistic regression with nested cross-validation and age- and sex-matched undersampling was used. Mood states were classified as depressive state (DS; Hamilton Depression Rating Scale-17 [HAMD-17] ≥8, Young Mania Rating Scale [YMRS] ≤7), state with manic/mixed symptoms (MS/MX; YMRS ≥8), or euthymic state (ES; HAMD-17 ≤7, YMRS ≤7). Seven classification tasks included BP vs HC, each mood state vs HC, and interstate comparisons. Feature importance was validated via permutation testing (P<.05). Medication sensitivity analyses were performed.

Results: The model achieved high discrimination for BP vs HC (area under the receiver operating characteristic curve [AUC]=0.916, 95% CI 0.854‐0.958; sensitivity 0.933; specificity 0.859) and strong performance for mood states vs HC (AUC=0.805‐0.874). Acoustic features comprised 96% to 98% of the features achieving statistical significance (P<.05), including mel-frequency cepstral coefficients, spectral characteristics, and voice quality measures, showing strong correlations with diagnostic probabilities (|ρ|=0.35‐0.85) and symptom severity (HAMD-17: ρ=–0.51; YMRS: ρ=–0.37); 2 to 4 linguistic features per comparison also reached significance. With sample independence ensured for interstate comparisons, DS vs ES significantly exceeded chance (AUC=0.760, permutation P=.03), whereas MS/MX vs ES (AUC=0.601, P=.20) and DS vs MS/MX (AUC=0.569, P=.30) did not, suggesting that markers primarily reflect persistent disease characteristics with partial state-associated components in depressive states.

Conclusions: The combined acoustic-linguistic model, driven primarily by acoustic features, shows potential as an objective aid for BP diagnosis (AUC>0.9), largely free of medication confounding; some features remained detectable during euthymia, although ES discrimination could not be fully separated from concurrent antipsychotic treatment. Although DS showed partial distinguishability from ES, overall limitations remain for monitoring mood state changes. Findings require validation in diverse populations and against other psychiatric conditions.

Related

  1. Predicting One-Year Psychiatric Rehospitalization Using Natural Language Processing and Machine Learning Methods Based on Clinical Texts
    Sare Aydın, Sena Karaca, Esma Akpınar Aslan +4
    Journal of clinical medicinePeer-reviewedCC BYFull textGeneral medicineEurope PMC
  2. Incorporating natural language processing-derived features from pathology reports into machine learning models to predict differentiated thyroid cancer recurrence
    Denise T. Lee, Steven Yen, Grace Deng +6
    Frontiers in EndocrinologyPeer-reviewedCC BY 4.0Full textEndocrinology & metabolismCrossrefEurope PMCPMC
  3. Do thoughts actually drive symptoms? Testing cognitive theories of posttraumatic stress with natural language processing in daily diaries
    Whitney R. Ringwald, Vasudha Varadarajan, Scott Feltman +6
    PsyArXivPreprintAll rights reservedFull textPsychiatry & mental healthOSF
  4. Natural Language Processing for Risk Stratification of Emergency Department Patients with Chest Pain
    Reza Valimoradi, George Cao, Allen Ding +9
    CJC openPeer-reviewedCC BY-NC-NDFull textCardiovascular medicineEurope PMC
  5. Combining natural language processing with patient-reported outcome measures scores to investigate the association of pandemic regulations with anxiety in children
    Michiel A. J. Luijten, Chris Gibbons, Conrad J. Harrison +8
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textPsychiatry & mental healthCrossref

Cite

BibTeXRIS
BibTeX
@article{yamaguchi2026machine,
  title = {{Machine Learning Classification of Bipolar Disorder Across Mood States Using Natural Language Processing and Speech Analysis: Observational Study}},
  author = {Yusei Yamaguchi and Yoshinobu Kano and Genichi Sugihara and Ryo Takemura and Hironobu Nakamura and Masaaki Shimizu and Mari Iizuka and Momoko Kitazawa and Ayako Sento and Hidehiko Takahashi and Taishiro Kishimoto},
  journal = {Journal of Medical Internet Research},
  year = {2026},
  month = nov,
  doi = {10.2196/98094},
  url = {https://doi.org/10.2196/98094},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Machine Learning Classification of Bipolar Disorder Across Mood States Using Natural Language Processing and Speech Analysis: Observational Study
AU  - Yusei Yamaguchi
AU  - Yoshinobu Kano
AU  - Genichi Sugihara
AU  - Ryo Takemura
AU  - Hironobu Nakamura
AU  - Masaaki Shimizu
AU  - Mari Iizuka
AU  - Momoko Kitazawa
AU  - Ayako Sento
AU  - Hidehiko Takahashi
AU  - Taishiro Kishimoto
PY  - 2026
DA  - 2026/11/01
JO  - Journal of Medical Internet Research
DO  - 10.2196/98094
UR  - https://doi.org/10.2196/98094
AB  - Background: Bipolar disorder (BP) affects 1% to 3% of adults worldwide, with diagnostic delays averaging 6 years and high initial misdiagnosis rates. The absence of objective diagnostic measures contributes to clinical uncertainty and suboptimal outcomes.  Objective: This study aimed to (1) develop a machine learning model for BP diagnosis and mood state classification using linguistic and acoustic features, (2) identify features associated with different mood states by comparing each with healthy controls (HC), and (3) explore whether features distinguish between mood states within BP.  Methods: We analyzed data prospectively collected over 7 years from 61 patients with BP and 99 HC across Japan. Speech samples were collected via semistructured interviews over 1 to 5 visits and analyzed cross-sectionally, using aggregated or single-visit observations per participant rather than within-person longitudinal modeling. We extracted 6373 acoustic features (openSMILE 3.0, ComParE 2016) and 211 linguistic features. Least absolute shrinkage and selection operator (LASSO)–regularized logistic regression with nested cross-validation and age- and sex-matched undersampling was used. Mood states were classified as depressive state (DS; Hamilton Depression Rating Scale-17 [HAMD-17] ≥8, Young Mania Rating Scale [YMRS] ≤7), state with manic/mixed symptoms (MS/MX; YMRS ≥8), or euthymic state (ES; HAMD-17 ≤7, YMRS ≤7). Seven classification tasks included BP vs HC, each mood state vs HC, and interstate comparisons. Feature importance was validated via permutation testing (P<.05). Medication sensitivity analyses were performed.  Results: The model achieved high discrimination for BP vs HC (area under the receiver operating characteristic curve [AUC]=0.916, 95% CI 0.854‐0.958; sensitivity 0.933; specificity 0.859) and strong performance for mood states vs HC (AUC=0.805‐0.874). Acoustic features comprised 96% to 98% of the features achieving statistical significance (P<.05), including mel-frequency cepstral coefficients, spectral characteristics, and voice quality measures, showing strong correlations with diagnostic probabilities (|ρ|=0.35‐0.85) and symptom severity (HAMD-17: ρ=–0.51; YMRS: ρ=–0.37); 2 to 4 linguistic features per comparison also reached significance. With sample independence ensured for interstate comparisons, DS vs ES significantly exceeded chance (AUC=0.760, permutation P=.03), whereas MS/MX vs ES (AUC=0.601, P=.20) and DS vs MS/MX (AUC=0.569, P=.30) did not, suggesting that markers primarily reflect persistent disease characteristics with partial state-associated components in depressive states.  Conclusions: The combined acoustic-linguistic model, driven primarily by acoustic features, shows potential as an objective aid for BP diagnosis (AUC>0.9), largely free of medication confounding; some features remained detectable during euthymia, although ES discrimination could not be fully separated from concurrent antipsychotic treatment. Although DS showed partial distinguishability from ES, overall limitations remain for monitoring mood state changes. Findings require validation in diverse populations and against other psychiatric conditions.
SN  - 1439-4456
SN  - 1438-8871
ER  -