Digital Musculoskeletal Assessment: A Narrative Review of Self-Report, Telehealth, Sensor-Based, and Conversational AI Modalities, with Implications for Workplace Screening

AbstractBackground:Digital technologies are increasingly proposed as scalable alternatives to in-personmusculoskeletal (MSK) assessment, spanning selfreport symptom screening, synchronous telehealth,wearable/sensor-based measurement, and, most recently, conversational artificial intelligence (AI). Thesefour modalities are often discussed interchangeablydespite measuring fundamentally different constructs,and interest in their workplace application — for earlyscreening and prevention rather than diagnosis — isgrowing faster than the underlying evidence base.Objective: To synthesise the highest-quality evidenceavailable for each modality, clarify what each is actuallyvalidated to do, and assess the extent to which evidencegenerated largely in symptomatic clinical populationscan inform deployment in occupational screening andprevention contexts.Methods: Narrative review structured according toSANRA criteria, prioritising systematic reviews andmeta-analyses, and supplemented with primary studieswhere no synthesis existed.Results: Telehealth diagnostic concordance is themost robustly evidenced modality, anchored by a registered meta-analysis (κ=0.80), though agreement variessubstantially by anatomical region. Digital self-reportscreening inherits strong paper-based psychometricvalidation, and GRADE-rated evidence supports highcertainty for digitally administered patient-reportedoutcome measures and pain assessment specifically— though direct digital validation of individual MSKspecific instruments (e.g., the Örebro questionnaire) remains largely absent. Sensor-based and wearable technologies are technically mature, achieving high classification accuracy, but are validated almost exclusivelyagainst conventional risk-classification proxies (RULA,REBA, NIOSH) rather than diagnosed MSK outcomes;where the underlying exposure-pain association hasbeen tested directly within this literature, it has notbeen confirmed, and the broader evidence on postural/biomechanical exposure as a predictor of pain —particularly static exposure in sedentary populations —remains contested, in contrast to more consistent evidence for repetitive and combined dynamic exposures.Conversational AI is the least mature modality, supported only by recent primary studies with no availablesystematic synthesis on the patient-facing symptomassessment task specifically.Conclusions: Evidence quality across the four modalities is uneven, and in several places thinner thancommonly assumed. A coherent digital MSK assessment pathway combining these modalities for workplace deployment remains, on present evidence, an aspiration rather than a validated system. Priorities for future research include direct digital validation of established screening instruments, outcome-linked (ratherthan proxy-linked) validation of sensor-based tools, andsystematic evaluation of conversational AI for patientfacing MSK symptom assessment specifically.Keywords: musculoskeletal disorders; occupationalhealth; telehealth; wearable sensors; digital health; artificial intelligence; screening