What happens between follow-up visits? The rise of fracture-healing sensors

Fracture healing is continuous, but clinical assessment remains episodic. Implantable sensors, wearable devices, and intelligent data analysis may offer a new way to follow recovery, but measuring a signal is only the first step toward improving clinical decisions. 

At a routine follow-up visit, an orthopedic surgeon may compare a new X-ray with an image taken several weeks earlier, ask about pain, examine the fracture site, and assess whether the patient can bear weight. These are essential parts of fracture care, but they leave an important question unanswered: What has been happening between visits?

Fracture healing is not a series of isolated events. It is a continuous biological and mechanical process in which inflammation, callus formation, mineralization, remodeling, and progressive restoration of load-bearing capacity overlap over time. Yet the tools used to monitor healing provide only occasional snapshots.

This mismatch can have an impact on recovery and healing. Decisions about advancing weight bearing, adjusting rehabilitation, dynamizing or removing an implant, ordering additional imaging, or investigating delayed union are often made without an objective longitudinal measure of fracture stability.

Sensor-based monitoring could help close this gap. New implantable, implant-adjacent, and wearable technologies are opening the possibility of moving fracture follow-up beyond occasional clinical snapshots toward continuous, quantitative, and data-informed assessment. Together, they provide three complementary windows into recovery: mechanical competence, local biological progression, and functional recovery. 

The limitations of the current snapshot

Plain radiographs remain central to fracture follow-up because they are accessible, familiar, and relatively inexpensive. They can show callus formation, cortical bridging, alignment, and changes in the fracture line.

However, radiographic signs of union largely reflect mineralized tissue that has already formed. They may therefore lag behind the underlying biological and mechanical changes occurring at the fracture site. More advanced modalities, such as CT and MRI, provide additional information but are not practical for continuous or frequent monitoring because of cost, radiation exposure, availability, and workflow demands.

Clinical findings also have limitations. Pain, tenderness, and weight-bearing ability are influenced by factors beyond fracture stability, including individual pain perception, muscle weakness, fear of movement, rehabilitation adherence, and comorbidities. Even experienced clinicians may find it difficult to estimate fracture stiffness objectively from examination and imaging alone.

The problem is therefore not that clinicians lack information. The problem is that most available information is intermittent, indirect, and weakly connected to the changing mechanical competence of the healing fracture. 

Three windows into fracture recovery

No single sensor can capture the entire healing process. A mechanical signal cannot fully describe tissue biology, and a wearable activity signal cannot confirm structural union. A useful way to understand the field is to consider three distinct but complementary monitoring strategies. 

1. Mechanical monitoring: Who is carrying the load?

Following fracture fixation, the healing bone and the fixation construct share physiological loads. During early healing, the fracture site is relatively compliant, so the plate, nail, or external fixator carries a greater proportion of the load. As callus forms and the fracture becomes stiffer, more load is transferred back to the bone. The mechanical demand on the fixation construct should therefore decrease.

Sensors can detect this transition by measuring implant strain, bending, deflection, compliance, or changes in resonant frequency. In principle, these signals provide a more direct connection to the mechanical endpoint of healing than radiographic appearance alone. They may eventually help inform decisions about weight-bearing progression, implant dynamization, or the need for closer investigation.

Early human studies have already demonstrated the feasibility of this concept. Instrumented femoral internal fixators used by Kienast and colleagues and Seide and colleagues showed progressive reductions in implant elasticity during healing. The longitudinal patterns could distinguish faster healing from slower healing and stable recovery from persistent instability. These studies remain among the clearest demonstrations that implant-load trajectories can provide clinically interpretable information over time.

External fixation systems have also been equipped with strain gauges, wireless communication, and remote data collection. In one clinical feasibility study, implant loading was monitored for an average of 139 days, showing characteristic changes as load progressively transferred from the fixation device to the healing limb. 

The AO Fracture Monitor as a translational example

A particularly relevant example is the AO Fracture Monitor T1. The system uses a battery-powered implantable data logger mounted on a compatible femoral locking compression plate. Strain-based measurements detect deformation of the plate during physiological loading, and implant-load data are transmitted by Bluetooth to a smartphone and cloud-based infrastructure.

Preclinical evaluation through the Smart Plate project demonstrated the feasibility of long-duration implant-load monitoring during fracture healing. The technology had progressed to a prospective first-in-human investigation in patients with femoral fractures or femoral instability treated using a compatible locking compression plate. The initial clinical study was designed primarily to evaluate safety, with clinical, radiographic, and technical performance included as secondary outcomes.

The significance of this project is not that it can already provide a definitive digital diagnosis of union. Validated thresholds for declaring healing, advancing weight bearing, or recommending intervention have not yet been established. Its significance is translational: it takes implant-load monitoring beyond an experimental engineering concept and moves it toward evaluation under real clinical conditions. It may help determine whether longitudinal mechanical information can become practical, reliable, and meaningful in routine fracture care. 

Battery-free wireless sensing

A different emerging approach is the development of battery-free implantable sensors. Instead of depending on an internal battery, these systems receive power wirelessly from an external wearable device. Near-field communication can be used both to energize the implanted sensor and to transfer measurements.

Kasper and colleagues developed battery-free, bone-mounted sensors containing metal-foil strain gauges. The implanted electronics were powered and interrogated through a wearable biosymbiotic hub. In large-animal fracture-healing studies lasting up to 11 months, the system demonstrated long-term operation and recorded progressive changes in bone strain adjacent to the healing fracture.

The researchers identified a characteristic strain pattern that corresponded with callus formation and other assessments of healing. This work demonstrates the possibility of monitoring fracture recovery without relying on a conventional implanted battery.

Battery-free systems may avoid limitations associated with battery size, depletion, replacement, and long-term storage. However, they still require an external wearable component for power and communication. Reliable coupling must also be maintained through skin, soft tissue, movement, and changing body position.

The current evidence remains preclinical. Human studies are needed to determine whether battery-free systems can provide stable, interpretable, and clinically useful measurements during routine fracture recovery. 

2. Biological monitoring: What is happening inside the callus?

Mechanical monitoring assesses how load is shared, but biological sensing aims to detect changes in the tissue itself. During secondary fracture healing, the local environment progresses through hematoma, inflammatory tissue, cartilage, immature bone, and remodeled bone. These tissues differ in water content, cellular organization, mineralization, electrical conductivity, and dielectric properties.

Electrical impedance spectroscopy is one approach to measuring these changes. Small electrodes apply alternating electrical signals across a range of frequencies and measure how the tissue resists or reacts to current. Changes in resistance, reactance, and impedance may provide information about the evolving composition of the fracture callus.

In mouse fracture models, impedance measurements increased as bridging callus formed and correlated with micro-CT measures of bone volume and mineral density. In a rabbit tibial model, a wireless plate-integrated impedance system recorded measurements for up to 10 weeks. Impedance followed a characteristic rise as the local tissue progressed from hematoma and cartilage toward bone and then approached a plateau during consolidation.

This creates an attractive possibility: biological sensing may detect changes before mineralized callus becomes clearly visible on radiographs. However, impedance is not a direct measurement of union. Electrode position, temperature, hydration, inflammation, edema, nearby metallic hardware, fibrous encapsulation, protein adsorption, and long-term changes at the sensor–tissue interface can all influence the signal. A change caused by sensor drift or local fluid variation could resemble a change caused by callus maturation.

For this reason, tissue-state sensors should currently be considered potential biological trend or risk-stratification tools rather than independent indicators of fracture stability, complete union, or readiness for unrestricted weight bearing. 

3. Functional monitoring: How is the patient using the limb?

Implantable sensors focus on the fracture site or fixation construct. Wearable devices provide a different perspective: how recovery affects the patient’s behavior and function in daily life. Pressure-sensing insoles, inertial measurement units, step counters, and activity monitors can quantify gait symmetry, loading distribution, activity levels, and use of the injured limb. Unlike a short walking assessment in a clinic, wearable monitoring may capture what happens at home, at work, and during rehabilitation.

One study by Chen and colleagues combined plantar-pressure measurements with a machine-learning model to estimate modified Radiographic Union Score for Tibial fractures, or mRUST, in patients treated with intramedullary nails. The model achieved a mean absolute error of 0.641 and an R² of 0.902. Explainability analysis also identified changes in the plantar regions contributing most strongly to predictions at different stages of recovery.

This does not mean that an insole can independently confirm union. Gait and limb loading are affected by pain, muscle strength, fear avoidance, rehabilitation instructions, age, neurological conditions, and other comorbidities.

The most appropriate role for wearable monitoring may be trajectory surveillance. A patient whose activity, gait symmetry, or limb loading stops improving—or changes unexpectedly—could be identified for earlier clinical review, imaging, or mechanical assessment. Wearables may therefore be particularly valuable for triage, remote follow-up, and rehabilitation management. 

The real challenge: Turning signals into decisions

Developing a sensor that produces a stable signal is a major engineering achievement. It is not, however, the final clinical objective. Clinicians do not simply need more strain curves, impedance values, step counts, or pressure maps. They need interpretable answers:

• Is this patient recovering along the expected trajectory?
• Is the fracture becoming mechanically stronger?
• Is the current loading level appropriate?
• Is an abnormal pattern developing?
• Should the patient return for imaging or examination earlier than planned?

Raw sensor data can be affected by many factors. Implant strain may change because the fracture has stiffened, but it may also decline because the patient has become less active. Muscle contraction, body weight, gait, fixation geometry, soft-tissue damping, and the loading task can all influence mechanical signals.

A useful system must therefore combine sensor measurements with context. At a minimum, this may include standardized loading tasks, patient activity levels, fracture type, fixation configuration, time from surgery, clinical findings, and imaging results. The central translational challenge is to transform a continuous signal into a clinically meaningful trajectory with a known degree of uncertainty. 

Where artificial intelligence may help

Artificial intelligence and machine learning may be useful for analyzing complex longitudinal data. Instead of relying on one measurement or one fixed threshold, a model could learn an expected recovery trajectory and identify meaningful deviations.

Future systems could integrate implant load, local tissue measurements, wearable activity, radiographs, patient characteristics, and comorbidities. This could support individualized estimates of healing progression and identify patients at increased risk of delayed union.

However, artificial intelligence should remain a clinical decision-support tool. It should not become an autonomous judge of whether a fracture has united or whether a patient may advance weight bearing. The quality of any model will depend on the quality, representativeness, and clinical relevance of the data used to develop it. 

A realistic path to clinical adoption

The immediate goal should not be to replace radiographs or clinical assessment. A more realistic near-term model is complementary monitoring. Mechanical sensors may provide information about load transfer. Biological sensors may indicate changes in local tissue state. Wearables may show whether functional recovery is following the expected trajectory. When combined, these data streams may offer a more complete picture than any one modality alone.

Translation should proceed in stages. First-in-human studies should establish safety, usability, signal stability, data completeness, and patient acceptance. Observational studies should then examine how sensor trajectories relate to radiographic union, mechanical stability, complications, and patient-centered outcomes.

Only after these relationships have been demonstrated should clinical trials test whether sensor-informed care improves decisions, reduces unnecessary imaging or visits, identifies delayed healing earlier, or improves functional recovery.

Clinical integration will also require attention to regulation, cybersecurity, privacy, interoperability with electronic health records, cost-effectiveness, and equitable access. A monitoring pathway that depends entirely on a modern smartphone and stable internet connection could unintentionally exclude some patients. 

From more data to better fracture care

The future of fracture-healing monitoring will not be determined by the number of measurements a device can collect. It will be determined by whether those measurements help clinicians and patients make better decisions.

Load-path mechanical monitoring currently has the strongest translational foundation because it is closely related to restoration of mechanical competence. Biological sensing may provide earlier insight into callus progression, while wearable devices offer a scalable way to monitor real-world recovery. None is sufficient alone.

The most promising future may lie in carefully validated, multimodal systems that combine these complementary perspectives. Fracture healing is continuous. Our monitoring strategies should eventually become capable of following that continuity—not to replace clinical judgment, but to make it more objective, timely, and individualized.