Frontiers in medicine · 2026
Conversational speech for respiratory triage in primary care: a pilot study.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Ravi Vijay, Noufi Camille
Аннотация
BACKGROUND: Respiratory complaints account for a substantial share of adult ambulatory visits, and accurate triage has direct consequences for antibiotic stewardship and pathogen-specific therapy. Prior work has investigated voice as a triage signal, but that literature is dominated by single-condition detection from scripted speech in crowdsourced or controlled clinical settings and has not been evaluated at the primary care scale using conversational ambient audio.
METHODS: A dataset of 514,377 ambient-recorded primary care visits from 379,225 adult patients at a US clinic network was used, with per-visit clinically assigned ICD-10 diagnosis codes and de-identified demographic and geographic metadata. Patient audio was extracted from each doctor-patient conversation, and spectral, voice quality, and prosodic features were computed. Eleven binary classification tasks were defined, aligned with a respiratory triage cascade (e.g., acute respiratory vs. acute non-respiratory illness, and lower vs. upper respiratory tract infection). An acoustic model was trained independently for each task using patient-stratified 5-fold cross-validation and evaluated on a held-out test set. Each model was also compared against six non-acoustic baselines using a single demographic, geographic, or temporal variable. The 11 trained classifiers were combined into a hierarchical cascade and illustrated as case studies.
RESULTS: Test-set AUC across the 11 tasks ranged from 0.602 (95% CI: 0.588-0.614) to 0.745 (95% CI: 0.742-0.748), with a mean expected calibration error of 0.018. After multiple-testing correction, six of the eleven binaries outperformed all six confounder baselines. Four binaries showed a median within-stratum AUC of 0.61-0.70 when the confounder was held fixed, indicating acoustic discrimination beyond what the confounder alone explains. Five binaries failed at least one axis; the only one outperformed by a confounder baseline was the pneumonia vs. non-pneumonia lower respiratory tract infection binary, which failed against the patient-city confounder baseline, plausibly reflecting a clinic-level difference in ICD-10 coding.
CONCLUSION: Conversational primary care audio contains an acoustic signal that discriminates clinically meaningful respiratory contrasts. Absolute performance is moderate, but the conditions are stricter than in prior work: conversational speech and differential-diagnosis contrasts among patients with illness. This pilot study establishes a baseline for voice-based clinical AI, advancing from sick-vs.-healthy detection toward differential-diagnosis panels and demonstrating a proof of concept for hierarchical composition.
METHODS: A dataset of 514,377 ambient-recorded primary care visits from 379,225 adult patients at a US clinic network was used, with per-visit clinically assigned ICD-10 diagnosis codes and de-identified demographic and geographic metadata. Patient audio was extracted from each doctor-patient conversation, and spectral, voice quality, and prosodic features were computed. Eleven binary classification tasks were defined, aligned with a respiratory triage cascade (e.g., acute respiratory vs. acute non-respiratory illness, and lower vs. upper respiratory tract infection). An acoustic model was trained independently for each task using patient-stratified 5-fold cross-validation and evaluated on a held-out test set. Each model was also compared against six non-acoustic baselines using a single demographic, geographic, or temporal variable. The 11 trained classifiers were combined into a hierarchical cascade and illustrated as case studies.
RESULTS: Test-set AUC across the 11 tasks ranged from 0.602 (95% CI: 0.588-0.614) to 0.745 (95% CI: 0.742-0.748), with a mean expected calibration error of 0.018. After multiple-testing correction, six of the eleven binaries outperformed all six confounder baselines. Four binaries showed a median within-stratum AUC of 0.61-0.70 when the confounder was held fixed, indicating acoustic discrimination beyond what the confounder alone explains. Five binaries failed at least one axis; the only one outperformed by a confounder baseline was the pneumonia vs. non-pneumonia lower respiratory tract infection binary, which failed against the patient-city confounder baseline, plausibly reflecting a clinic-level difference in ICD-10 coding.
CONCLUSION: Conversational primary care audio contains an acoustic signal that discriminates clinically meaningful respiratory contrasts. Absolute performance is moderate, but the conditions are stricter than in prior work: conversational speech and differential-diagnosis contrasts among patients with illness. This pilot study establishes a baseline for voice-based clinical AI, advancing from sick-vs.-healthy detection toward differential-diagnosis panels and demonstrating a proof of concept for hierarchical composition.