Digital Physiotherapy in Orofacial Pain

Digital physiotherapy can help deliver education, supervised exercise and follow-up for people with temporomandibular disorders (TMD). Its value depends on the treatment delivered and the patient’s response. A video appointment, a therapeutic app, an electromyography sensor and an artificial-intelligence model perform different jobs; evidence for one does not establish the effectiveness of the others.

This article focuses on rehabilitation for assessed TMD-related pain and jaw-function limitations. Orofacial pain has other possible causes, so choosing the right diagnosis comes before choosing a platform.

Start with the problem the patient wants to solve

Ask what hurts, when it occurs, how long it has persisted and which activities are difficult. Record jaw function alongside pain, daily interference and relevant psychosocial factors. The Diagnostic Criteria for TMD combine a clinical diagnostic assessment with measures of pain-related disability, distress and function. A digital record can help organize this information; a contact map or muscle trace cannot replace that assessment.

For a patient who finds chewing uncomfortable, the useful baseline is the symptom and functional limitation to be reviewed later. More recorded signals are worthwhile only when they answer a clinical question. This keeps rehabilitation focused on the person’s daily activities rather than achieving an attractive dashboard.

What can be delivered digitally?

Education, exercise instruction, symptom recording and clinician feedback are practical components of a digital pathway. For adults with moderate chronic TMD pain lasting at least three months, the 2023 BMJ guideline favours interventions including supervised jaw exercise, stretching, therapist-assisted mobilization, usual care and cognitive behavioural therapy (CBT). These recommendations concern the interventions; they do not establish that every app delivers them effectively. The guideline does not address acute TMD pain.

Two trials illustrate the distinction:

• Remote supervision: Abaci and colleagues randomized 40 adults with TMD and probable sleep bruxism to education and home exercises with or without weekly video supervision for eight weeks. Both groups improved; supervision produced greater improvements in morning jaw pain and several other measured outcomes, while sleep-quality improvement was similar. The study was conducted during the COVID-19 pandemic, which limits generalization. It tested supervised exercises, not autonomous AI treatment.
• A specific therapeutic app: Park and colleagues randomized 102 adults to a six-week multicomponent TMD app or a symptom-recording sham app. The therapeutic program combined education, jaw exercises, behavioural content, meditation and feedback. Both groups continued usual treatment. Its primary per-protocol analysis included 93 participants and favoured the therapeutic app for pain; low-adherence participants were excluded. Short follow-up limits conclusions about durability, and developer involvement and company funding were disclosed. The combined program’s result cannot identify which component caused the benefit or validate unrelated apps.

When selecting a service, therefore, ask what the patient will actually receive, who will review progress and how the exercises will be adapted. Check that the patient can use the platform comfortably, and agree on an alternative when access, confidence with technology or symptoms make remote care unsuitable.

Use EMG feedback to support learning, not to define recovery

Surface electromyography (sEMG) records electrical activity associated with muscle activation. A biofeedback system presents a measured signal so a person can practise changing a response. NIDCR describes jaw-muscle feedback as one way a therapist may help someone notice tension and practise relaxation. A wearable format makes a sensor portable; it does not establish that the system diagnoses TMD, measures pain or chooses an effective treatment.

The clinical context matters. The BMJ guideline favours CBT with or without biofeedback, but makes a conditional recommendation against biofeedback used on its own for the chronic-pain population above. Combining these categories into a general claim that “EMG biofeedback works” would obscure the difference.

In a randomized trial of 58 people with chronic TMD, eight weeks of biofeedback-based CBT and eight weeks of splint treatment produced similar clinically meaningful improvements in pain and disability. CBT participants improved more in pain coping; nocturnal masseter activity did not change significantly. Clinical improvement and a lower muscle-activity reading are therefore different outcomes. This comparison does not establish a universal indication for splints.

For repeated recordings, document the device, recording task and relevant acquisition conditions so the clinician can judge whether observations are comparable. Review the trace alongside the patient’s pain, function and experience. A changed trace is a reason to interpret the data, not an automatic instruction to intensify exercises or stimulation.

Where do occlusal records and AI belong?

Digital impressions describe dental surfaces and recorded tooth relationships. Contact-analysis systems and sEMG add different measurements. None of those records, by itself, demonstrates that a bite discrepancy caused the pain. NIDCR reports that current evidence does not support a strong link between malocclusion and TMD; there is insufficient evidence for bite-changing treatments as TMD therapy, and they can worsen symptoms. Braces, crowns or tooth grinding should not follow simply from a rehabilitation dashboard. A separate dental or restorative indication needs its own assessment.

Likewise, software that displays a signal is not necessarily AI. When a platform claims to predict recovery or automatically personalize treatment, ask for validation of that exact system in a comparable TMD population. Did the study measure patient pain and function, or only classify signals? Was it compared with an appropriate control? Did a clinician supervise changes? The trials described here support specific supervised or multicomponent approaches; they do not establish autonomous treatment decisions from combined bite and EMG data.

TENS is also a separate intervention from recording EMG. NIDCR notes that TENS studies for TMD have been small and short in duration; the BMJ guideline conditionally recommends against it for the chronic-pain population described above. A sensor-equipped platform does not remove those evidence limits.

Bring the information back to a clinical decision

Agree on a review point and a small set of patient-relevant outcomes before starting. Reassess the original activity limitation, pain and ability to follow the plan, along with adverse effects and any useful recordings. If symptoms worsen or the agreed goals are not being met, review the diagnosis, exercise delivery and need for in-person or specialist assessment rather than escalating automatically.

The most useful digital pathway connects an assessed problem, a suitable conservative intervention and a meaningful review. Clinicians can use technology to maintain contact and make progress easier to discuss while keeping clinical decisions grounded in the patient’s response.