Precision acupuncture: Moving from art to evidence-based science

What constitutes an “adequate dose” of acupuncture remains an important question in acupuncture research. White and colleagues proposed a neurophysiological framework in which dose is defined by the totality of physical procedures applied in each session and the patient’s resulting responses. ¹ Subsequent work has extended this concept to a multidimensional construct encompassing needle number, session frequency, the total number of treatment sessions and, critically, stimulation intensity. ² Nevertheless, in both research and practice, these parameters remain inconsistently specified, which limits reproducibility, training quality and the interpretability of clinical trials.

Recent technological advances are moving acupuncture from an empirically guided art toward a more measurable and standardized intervention. Three domains are central to this shift: (1) intensity-centered dose–response relationships; (2) three-dimensional (3D) anatomical targeting and needling depth; and (3) technology-assisted quantification and education of manipulation skills.

First, a systematic review has suggested that stimulation intensity—operationalized through the amplitude and frequency of needle manipulation, insertion depth and retention time—may exert a greater influence on clinical outcomes than needle number or session frequency in conditions such as dysmenorrhea, knee osteoarthritis and musculoskeletal pain. ² At the same time, emerging mechanistic studies suggest that stronger stimulation is not always superior, highlighting the need for site-specific and condition-specific optimization of stimulation parameters. Tonification and sedation can be reinterpreted as intensity levels rather than inherently distinct techniques: weaker stimulation for deficiency patterns, stronger for excess states, with intensity titrated according to individual responses. In this context, de qi is better understood not as the dose itself, but as an immediate, clinically accessible response indicator that may reflect engagement of relevant afferent pathways and limbic–somatosensory networks. ³ The presence, quality and onset latency of de qi can guide real-time adjustment of intensity to achieve personalized yet standardized dosing.

Second, traditional acupuncture point locations should be approached as 3D anatomical targets rather than superficial locations. Needling depth determines which neuroanatomical structures—superficial nerve endings, deep somatic tissues (primarily muscle) or larger nerve bundles—are engaged, and therefore shapes both therapeutic effects and safety profiles. ⁴ Beyond depth, site-specific differences in connective tissue structure, innervation density and electrical properties may cause identical mechanical or electrical stimulation to produce different physiological effects across anatomical locations. High-resolution 3D point atlases integrating magnetic resonance imaging (MRI) and anatomical data now provide more precise guidance for depth adjustment across body regions and individual anatomical variaions. Ultrasound further improves precision by enabling real-time visualization of deep structures, supporting accurate needle placement and reducing adverse events in high‑risk regions such as GB21 and the upper abdomen. Together, these tools offer an anatomical foundation for reproducible, depth-aware dosing.

Third, the motor skills underlying acupuncture manipulation have traditionally been transferred through apprenticeship and subjective observation. Motion analysis now reveals a paradox: large inter‑individual variability in kinematic patterns, yet striking intra‑individual consistency—the practitioner’s unique but stable “signature” of amplitude, frequency and trajectory. ⁵ This finding underscores both the practitioner-dependent nature of acupuncture manipulation and the need for quantification. Three dimensional motion tracking, haptic training simulators with realistic needle grasp and sensor-based visual feedback platforms enable objective measurement of practitioner-specific manipulation patterns and structured training of novices. Artificial intelligence (AI)–driven systems can classify manipulation types and estimate stimulation parameters in real time, thereby enabling the development of standardized competency benchmarks and investigation of dose–response relationships in clinical practice.

These developments suggest several priorities for the acupuncture community. In clinical practice and research, acupuncture procedures should be routinely documented in terms of intensity-related parameters (manipulation amplitude, frequency and retention time), depth ranges and response markers such as de qi characteristics, in addition to needle number and session frequency. For clinical implementation, pragmatic studies are needed to evaluate whether ultrasound guidance and simulator-based training improve safety, effectiveness and cost-effectiveness, compared with conventional approaches. For personalizing dosing, emerging biomarkers—including gut microbiome profiles, neuroimaging signatures and inflammatory cytokines—should be evaluated to determine whether biomarker-guided dosing outperforms empirical titration guided by de qi alone.

In our view, the convergence of biomechanical metrics, advanced imaging and simulator- and AI-supported education offers an opportunity to redefine the concept of “adequate acupuncture.” By embracing intensity-centered dosing, depth-aware anatomy and technology-assisted training, the field can move toward safer, more standardized and evidence-based practice, alongside individualized clinical judgment.