Before the prescription: Could wearables help detect infection earlier and slow AMR?

What is the scale of the problem?

Antimicrobial resistance is an urgent global challenge – think just, 80 or so years ago, a bacterial infection could often be a death sentence. But then Fleming discovered penicillin and everything changed. However, even he had a stark warning.

In his 1945 Nobel lecture, Alexander Fleming warned:

“It is not difficult to make microbes resistant to penicillin.”

Ten years ago, the scale of the problem was brought into sharp focus by Jim O’Neill’s Tackling Drug-Resistant Infections Globally review:

“The magnitude of the problem is now accepted. We estimate that by 2050, 10 million lives a year and a cumulative 100 trillion USD of economic output are at risk due to the rise of drug resistant infections.”

Most recently, the 2024 Global Research on Antimicrobial Resistance (GRAM) study, published in The Lancet, forecast:

“Bacterial antimicrobial resistance will cause 39 million deaths between 2025 and 2050 — equating to three deaths every minute.”

What about the status quo?

Antimicrobial resistance (AMR) is usually discussed as a problem of drug discovery, prescribing behaviour and healthcare policy. All are critical. But there is another part of the challenge that receives less attention: timing. 

In many clinical settings, antibiotics are prescribed before the cause of infection is fully understood. This is often clinically necessary. A patient is unwell, the pathogen is unknown, and treatment cannot always wait for laboratory confirmation. The result is a system that often has to treat first and refine later. 

That uncertainty matters. When clinicians do not yet know whether an infection is bacterial, viral or inflammatory, broad-spectrum antibiotics can become the safest immediate option. In some cases, they are essential. In others, they may be unnecessary. The difficulty is knowing which is which early enough to act differently. 

This is why the next frontier in antimicrobial resistance may not only be faster laboratory testing. It may be earlier detection of physiological change, before a patient deteriorates, before they present to care, and before broad empirical antibiotics become the default option. 

That creates a new question for healthcare, diagnostics and MedTech innovators: could wearable technologies become part of the first line of defence against AMR?

AMR is also a timing problem

The current diagnostic pathway is often reactive. A patient develops symptoms, seeks care, provides a sample, and waits for a result. Even rapid diagnostics still usually depend on the patient entering the healthcare system at the right moment. 

You might be expecting me to talk about diagnostics and lateral flows here, but wearables create a different model. 

Instead of testing at a single point in time, wearable technologies can monitor continuous changes in physiology. Heart rate, heart rate variability, respiratory rate, temperature, sleep and activity can all shift when the body begins to respond to infection. These signals are not specific enough to diagnose infection on their own. But when measured continuously against a person’s own baseline, they may reveal early signs that something is changing. 

This reflects a key shift embedded within the NHS 10 Year Health Plan. The value of wearables is not in replacing diagnostic testing, but in enabling earlier identification of when testing, monitoring, or clinical review is required. More importantly, wearables can generate clinically relevant, longitudinal data that complements and strengthens traditional in vitro diagnostic (IVD) results obtained in clinical settings, such as blood or urine tests. 

The first wave: physiological signals and early illness detection

The first wave of infection-relevant wearable innovation has already happened. 

I don’t know about you, but often before I become unwell, I spot a change in myself. I recently had some food poisoning and not long after eating I noticed more wind than usual and a higher heart rate. I brushed it off as nothing but a few hours later, full blown symptoms appeared. For food poisoning there’s probably very little I could have done (unless it was a bacterial infection) but in other cases, if diagnostics or IVDs are available, you could identify the pathogen and intervene earlier. 

Interestingly during the COVID-19 pandemic, devices and platforms from Oura, Fitbit, Apple, Garmin and WHOOP were used in different studies or programmes exploring whether passively collected physiological data could identify changes linked to viral illness. 

Oura is a useful example because it has moved this idea into a consumer-facing product feature. Its Symptom Radar feature monitors changes in biometrics such as body temperature, respiratory rate, resting heart rate, heart rate variability and inactive time to identify signs of physiological strain. Oura is clear that this is not a medical diagnostic device. But it shows how wearable companies are starting to position continuous monitoring around early illness signals rather than only fitness or recovery. 

Scripps Research took a similar direction through its DETECT programme, using wearable data such as heart rate, sleep and activity to look for patterns associated with influenza, coronavirus and other viral illnesses. Apple Watch was also used in the Mount Sinai Warrior Watch Study, which explored whether heart rate variability could signal the onset of COVID-19 before diagnosis. WHOOP has been used in research examining whether changes in respiratory rate could act as a leading indicator of SARS-CoV‑2 infection. 

The important lesson from these examples is not that a fitness tracker can diagnose infection. It is that passive, continuous data can add useful context to symptom-based assessment. 

That distinction matters for AMR. The goal is not to claim that a watch can identify the pathogen. The value is in recognising that infection often changes physiology before a patient seeks care. If that change can be detected earlier, it may create an opportunity to test, monitor or intervene before broad-spectrum antibiotics become the default response. 

The second wave: clinical-grade remote monitoring

The next step is the movement from consumer wearables into clinical-grade monitoring. 

Platforms such as BioIntelliSense’s BioButton are designed for continuous patient monitoring and early identification of patient deterioration. Current Health uses an upper-arm wearable to passively monitor respiratory rate, heart rate, SpO2, skin temperature and activity. Empatica’s Health Monitoring Platform has received FDA clearance for continuous collection of physiological parameters including SpO2, skin temperature, activity and electrodermal activity, with later clearance for additional cardiac and respiratory digital biomarkers. 

These systems are not AMR diagnostics. But they are relevant to AMR because they show how continuous monitoring is being pulled into healthcare delivery. 

In hospitals, wearable monitoring could help detect deterioration between routine observations. In virtual wards and hospital-at-home models, it could support remote assessment of patients who might otherwise present late. In post-discharge care, it could help identify when a patient is beginning to deteriorate and needs review. In vulnerable groups, such as older adults, immunocompromised patients or people recovering from surgery, these earlier warning signals may be particularly valuable. 

For antimicrobial stewardship, the opportunity is to use continuous monitoring as a trigger for better decisions. That could mean earlier confirmatory testing, closer review, targeted diagnostics or more confidence in withholding antibiotics where watchful waiting is clinically appropriate. 

The third wave: moving beyond vital signs

The limitation of most current wearables is that they mainly detect indirect signs of illness.

A raised resting heart rate or change in sleep pattern can indicate infection, but it can also indicate stress, alcohol, poor recovery, travel, heat, medication or another underlying condition. Physiological data can show that something is changing, but it may not explain what is changing or why.

The more disruptive opportunity is to combine physiological monitoring with biochemical sensing.

Continuous glucose monitors from Dexcom and Abbott have already normalised the idea that a small skin-worn device can measure a biomarker in interstitial fluid over time. Dexcom’s systems measure glucose in interstitial fluid every few minutes. Abbott’s FreeStyle Libre sensors continuously measure glucose concentration in interstitial fluid. These products are focused on diabetes, not infection. But they have created an important proof point for the broader wearable diagnostics field: continuous biochemical monitoring can become mainstream when the use case is clear, the device is usable and the data supports practical decisions.

For AMR, the question is whether other biomarkers could follow a similar path.

If glucose can be measured continuously in interstitial fluid, what other clinically relevant signals might be measured in the future? Could wearable or minimally invasive platforms track inflammatory markers, immune response signatures, metabolic changes, tissue status or antibiotic concentrations? Could they help identify when infection risk is increasing, when treatment is working, or when a patient needs escalation?

Companies are already exploring parts of this landscape. Sava has reported clinical evidence for microsensor-based continuous glucose monitoring in interstitial fluid and describes a longer-term platform ambition for multi-molecule sensing. Nutromics is developing a microneedle-based Lab-on-a-Patch platform for continuous molecular monitoring, including work focused on therapeutic drug monitoring of vancomycin. That is particularly relevant to AMR because vancomycin is used in serious bacterial infections, and better real-time monitoring could support safer, more precise dosing. Another example is Epicore Biosystems who are taking a different route through sweat-sensing wearables, using microfluidics and biosensors to analyse biomarkers in perspired sweat.

These technologies are at different stages of maturity. hey do not show that a wearable AMR supporting diagnostic has arrived. They show that the building blocks are emerging.

Continuous physiological monitoring is becoming more established. Biochemical sensing is becoming more wearable. Remote care pathways are becoming more accepted. Digital tools are becoming more embedded in clinical monitoring.

In 2020, Dr Timothy Rawson of Imperial College London highlighted the potential of this technology:

“Microneedle biosensors hold a great potential for monitoring and treating the sickest of patients.”

The opportunity is to connect these developments into systems that support earlier, smarter infection management.

What would an AMR wearable actually need to do?

The value of a wearable infection technology will not be determined by the sensor alone. It will be determined by the decision it supports. 

For AMR, the most useful technologies may be those that help answer practical clinical questions earlier than current pathways allow: 

• Is this patient developing an infection? 

• Is this patient deteriorating? 

• Does this patient need a confirmatory diagnostic test? 

• Could antibiotics be safely avoided while monitoring continues? 

• Is antibiotic treatment working? 

• Is the dose within the right therapeutic range? 

• Should this patient be escalated before they become acutely unwell? 

A non-specific alert that increases anxiety, unnecessary testing or precautionary antibiotic prescribing could make the AMR problem worse. A validated system that identifies deterioration earlier, prompts targeted testing, supports triage or helps optimise treatment could become a valuable tool in antimicrobial stewardship. 

This distinction matters. The goal is not more data. The goal is better decisions. 

The future may be a hybrid pathway

The most credible future is unlikely to be a single wearable that diagnoses infection in isolation. A more realistic model is a hybrid pathway. A wearable may first detect deviation from a personal baseline.

As one senior UK-based pharma leader observed:

“What a patient needs to know is what is normal for them and when they are outside of their normal range, how to seek help.”

Building on that vision, AMR Insights wrote in 2019:

“Connected diagnostics, wearables for remote patient monitoring and smartphone capabilities potentially revolutionising the fight against AMR. Also think of future clinical development of novel antibiotics by point-of-care diagnostics and digital biomarkers to the patient.”

That signal may then trigger a digital assessment, a confirmatory near-patient test, remote clinical review, targeted sample collection or closer monitoring. In higher-risk groups, this could support earlier intervention before infection progresses.

In chronic wound care, a wearable or sensor-enabled dressing could help identify local or systemic signs of infection before the wound becomes severe. In post-operative care, continuous monitoring could identify patients who are deteriorating after discharge. In virtual wards, physiological and biochemical signals could support more precise escalation decisions. In hospitals, therapeutic drug monitoring could help clinicians optimise antibiotic dosing in patients with severe infections.

This is where the AMR opportunity becomes broader than diagnostics alone. It becomes a connected care challenge.

The innovation gap

Many promising diagnostic and wearable technologies fail not because the science is weak, but because translation is difficult. 

A next-generation infection-sensing wearable may need to combine biomarker biology, assay chemistry, skin-compatible materials, flexible electronics, microfluidics, microneedle interfaces, low-power data capture, digital algorithms, clinical validation, regulatory strategy and scalable manufacturing. 

Each element is challenging. The real difficulty is integrating them into a product that is reliable, manufacturable, acceptable to users and suitable for healthcare deployment. 

This is particularly important in AMR. The clinical consequences of getting the signal wrong can be significant. False reassurance could delay treatment. False alarms could increase unnecessary healthcare use or antibiotic prescribing. Poorly validated biomarkers could undermine trust. Devices that work in controlled studies may fail when used in real-world settings. 

For innovators, the challenge is therefore not simply proving that a signal can be measured. It is proving that the signal is robust, clinically meaningful and actionable. 

What this means for MedTech and diagnostic innovators

For companies developing wearables, biosensors, digital biomarkers or infection diagnostics, AMR represents a significant opportunity. But the market will not reward novelty alone. 

The most compelling technologies will be those that fit into real clinical pathways. They will need to work with existing diagnostic infrastructure, support clinical decision-making, generate evidence for adoption and meet regulatory expectations. They will also need to be manufacturable at a cost and scale that makes deployment realistic. 

This is where early design choices matter. Biomarker selection, sample access, materials, device format, usability, data capture, assay stability and manufacturing process all influence whether a promising prototype can become a deployable product. 

In other words, the future of wearable diagnostics for AMR will depend as much on translation as invention.

How CPI can help

CPI works with companies developing complex MedTech, diagnostic and wearable technologies, helping them move from concept towards commercialisation. 

For AMR-related diagnostics and infection monitoring, this can include support across wearable device design and development, in vitro diagnostic development, biosensors, assay integration, flexible and printed electronics, materials, adhesives, coatings, formulations, microfluidics, firmware, AI-enabled data interpretation, digital health integration, regulatory readiness, pilot production and manufacturing scale up. 

This combination is important because infection-sensing wearables sit between several disciplines. They are integrated medical technologies that require biology, chemistry, materials science, engineering, data and regulatory thinking to work together. 

CPI can help innovators address this complexity earlier in development, reducing technical risk and helping technologies move towards products that are reliable, scalable and clinically relevant. 

That means CPI can support innovators from early design choices through to scalable manufacturing: selecting skin-compatible materials, developing printed or flexible electronic components, integrating sensors and firmware, building AI-ready data workflows, and designing processes that can move beyond prototype devices into repeatable production. 

As AMR continues to challenge healthcare systems, earlier diagnostics will become increasingly important. The future of AMR may not begin only in the microbiology laboratory. It may begin with continuous monitoring, earlier warning, biochemical sensing and smarter decisions before the first prescription is written. 

That is the opportunity for the next generation of wearable diagnostics.