Artificial intelligence and osteopathic medicine: preserving human-centered care in an era of technological advancement

AI’s promise for osteopathic practice

Properly implemented, AI can enhance osteopathic practice, rather than replacing the physician’s role. Ambient scribing exemplifies this potential, utilizing audio and text-based systems to reduce documentation time (“pajama time”) [7] and minimize screen interaction. These tools return time to physicians for attentive listening, relational presence, and palpatory diagnosis – freeing hands for healing rather than typing.

AI excels in diagnostic support, processing diverse datasets – clinical histories, laboratory values, imaging, and patient-reported outcomes – to create comprehensive patient profiles that enhance care coordination [8]. This analytic capability enhances communication, care continuity, and incorporates social determinants of health [8].

Importantly, AI can identify subtle structure–function interdependencies that may be missed in time-constrained clinical encounters. Pattern recognition across complex physiological systems aligns with the osteopathic understanding of bodily interconnectedness [9].

By reducing the cognitive load of administrative tasks that add little clinical value, AI can help osteopathic physicians refocus on individual patients and their unique healing trajectories.

Personalized medicine

AI supports personalized medicine, complementing osteopathic individualized care and self-healing principles. Predictive analytics can guide interventions that resonate with each person’s unique presentation and healing capacity. AI-enhanced mind–body tools – such as wearable-guided breathing platforms (e.g., Breathwrk, Apollo Neuro) [10], 11] and real-time biofeedback devices (e.g., Muse, HeartMath) [12], 13] – extend psychosomatic care beyond clinic visits. These tools adapt to individual physiological patterns (heart rate variability, electroencephalography [EEG] signals) supporting autonomic balance and stress resilience while reinforcing mind-body-spirit unity.

Remote patient monitoring (RPM) improves patient mobility and functional status, and reduces hospitalizations and healthcare costs [14]. Although quality-of-life outcomes data (physical and mental health symptoms) remain inconclusive [14], RPM offers evidence-based tools aligned with osteopathic priorities like functional improvement. Continuous glucose monitoring (CGM) exemplifies this potential, improving glycemic control through precise and real-time monitoring within patients’ daily environments, enabling data-driven personalized feedback [15], [16], [17].

Erosion of clinical skills and diagnostic competence

Recent studies raise concerns about diagnostic performance when physicians lack adequate AI training [19], [20], [21], [22]. Meta-analyses reveal that while AI models like Generative Pre-trained Transformer 4 (GPT-4) can outperform conventional resources independently, they do not consistently improve diagnostic accuracy as decision-support tools [19], 22]. This paradox suggests that without proper training, physicians may over-rely on AI, potentially reducing rather than enhancing diagnostic accuracy.

Over-reliance on AI leads to cognitive atrophy and deskilling – diminished clinical reasoning skills, pattern recognition, and physical diagnostic skills. For osteopathic physicians, this erosion threatens more than medical judgment; it impacts core competencies like diagnostic palpation and OMT. Increased technological dependence reduces verbal and physical patient engagement, compromising the astute palpatory perception and OMT [5].

The cost of omission: missing language inputs and the displacement of osteopathic practice and sustainability

Current AI systems fundamentally mismatch with osteopathic philosophy and practice. Key elements of osteopathic care – somatic dysfunction, OMT, body-mind-spirit integration–are underrepresented in large language models and clinical algorithms due to limited osteopathic literature in training datasets. While PubMed contains over 38 million citations [23], osteopathic publications comprise only a small fraction. AI models trained on clinical documentation, coding data, and medical guidelines may not capture osteopathic terminology or practices, compounding this issue.

This underrepresentation creates risks: missed diagnostic opportunities, incomplete assessments, and inadequate documentation that fails to capture the full context of the healing process [2], 4]. AI-enabled ambient systems may miss osteopathic terminology or diagnostic codes, resulting in undercoding and inaccurate records that threaten care quality and reimbursement. Feeding osteopathic-specific language into AI clinical decision-support tools is fundamentally important for sustainable osteopathic practice.

Without intentional design integrating osteopathic language, diagnostic frameworks, and treatment modalities, AI risk marginalizes osteopathic perspectives. Developers and medical institutions must proactively preserve human-centered care and osteopathic principles in AI development and deployment [2], 4], 8].

Impact on therapeutic relationships and patient trust

The physician–patient relationship serves as a cornerstone of osteopathic care, yet increased reliance on AI-driven systems risks creating barriers to this therapeutic connection if thoughtful attention to preserving patient trust is overlooked during clinical support deployment. While ambient scribing and AI-assisted messaging reduces physician screen time, automation must enhance – not diminish – meaningful patient interaction.

Patients remain wary of AI’s role in clinical encounters. Recent findings reveal consistent anti-AI bias toward physicians disclosing any use of AI – whether administrative, diagnostic, or therapeutic [24]. Although modest, these effects carry potential clinical significance given the centrality of patient trust. Concerns include overreliance on algorithms, erosion of the personal touch, data privacy, and cost [24]. These fears underscore the importance of maintaining transparent, rational, and individualized care.

This matters given osteopathic medicine’s historical commitment to underserved communities. Although osteopathic physicians represent just 4.9 % of clinically active physicians, osteopathic physicians account for 10.4 % of rural primary care physicians and 15.3 % of physicians in small rural counties [25]. Here, osteopathic physicians critically maintain trust, support technology literacy, and ensure equitable access to care.

If AI tools exacerbate health inequities or erode patient autonomy, they may disproportionately harm the populations that are most dependent on osteopathic services [8], 18], 26]. Privacy concerns threaten therapeutic alliance when sensitive information is processed by opaque systems, causing patient hesitation to disclose emotional, social, or spiritual concerns [24], all of which are insights central to osteopathic assessment. Designing AI platforms without mechanisms to monitor and mitigate bias, protect patient privacy, and include osteopathic terminology risks perpetuating these harms to vulnerable communities [4], 8], 18], 26].

AI-generated recommendations built with limited inputs may exert undue influence on clinical decisions, pressuring patients toward standardized pathways that ignore social drivers of health critical to improved outcomes for patient populations served by osteopathic physicians.

Osteopathic-specific principles for AI integration

We propose these guidelines to ensure that AI integration aligns with the foundational tenets as outlined in the American Osteopathic Association’s Artificial Intelligence in Healthcare Report and Action Plan (Policy H429-A-24) [8]:

1. Representation requirement: AI systems must recognize osteopathic diagnostic terminology, OMT, and holistic clinical reasoning to prevent exclusion from AI-supported care pathways.

2. Skills preservation: Clinical decision support should enhance – not replace – osteopathic diagnostic and palpatory skills. AI systems should prompt relevant structural examinations and incorporate osteopathic-specific terminology and billing codes to accurately document and bill for services rendered.

3. Treatment parity: AI recommendation engines must include OMT as standard treatment when appropriate, rather than defaulting to pharmaceutical or surgical interventions.

4. Shared governance at institutional and national levels: Osteopathic physicians need voices in AI policy, development, and oversight to address bias and safeguard patient autonomy and privacy.

5. Defined scope of practice: AI platform deployment must include inputs allowing for the full scope of osteopathic care. Osteopathic-specific terminology and descriptors addressing somatic dysfunction and treatment modalities are critical for accurate coding, billing, and preservation of osteopathic medicine.

6. Shared decision-making: AI-supported care must prioritize shared decision-making between osteopathic physicians and patients, supporting precision medicine that is individually tailored to meet the patient’s needs within their environmental context.

7. Osteopathic-led innovation: Encourage osteopathic research on AI applications in OMT, psychosomatic medicine, and person-centered outcomes.

Educational imperatives for the AI era

Medical education must prepare future Doctors of Osteopathic Medicine (DOs) for AI-shaped healthcare while emphasizing palpatory ability, clinical reasoning, and empathic communication [5], 8], 27].

1. Dual literacy requirement: Students need proficiency in both traditional osteopathic diagnostic skills and AI-generated information interpretation.

2. Critical evaluation skills: Learners must critically appraise AI outputs, recognizing biases – such as prioritization of pharmacologic interventions over manual medicine – and identifying misalignment with osteopathic philosophy.

3. Professional identity formation: Curricula should emphasize how osteopathic principles remain essential in technology-enhanced environments.

Research priorities for osteopathic integration into AI

Osteopathic medicine must engage in targeted research to ensure that its values, methods, and outcomes are accurately reflected in AI systems.

1. Outcome studies: Demonstrating osteopathic intervention effectiveness utilizing AI-measurable outcomes (functional improvement, patient satisfaction, and reduced healthcare utilization) supports algorithm inclusion.

2. Diagnostic algorithm development: Translating osteopathic diagnostic reasoning into AI-suitable data formats ensures inclusion in future clinical tools.

3. Implementation research: Studies assessing how AI integration affects osteopathic clinical skills, patient care, and professional identity in real-world practice settings.