Across specialties and continents, researchers are working to unlock the diagnostic potential of perspiration.

And while sweat has long been understood as an accessible source of health information, there are signs its superpowers are about to be turbocharged.

Sweat testing has already proven its value in areas like cystic fibrosis diagnosis, while recent research has also shown it has promise for measuring glucose levels in those with diabetes [link here].

But as lecturer in analytical chemistry at University Technology Sydney, Dr Dayanne Bordin, and colleagues outlined in Journal of Pharmaceutical Analysis earlier this month [link here], advances in analytical techniques and artificial intelligence could further expand the role of sweat in health monitoring.

Evolutions in AI over the past fews years have opened the door for “improved pattern analysis and classification algorithms to improve diagnostic precision and therapeutic accuracy,” they said.

Specialists should view sweat as a non-invasive complementary diagnostic matrix offering real-time patient insights, Dr Bordin told the limbic. 

“Clinicians should anticipate sweat-based diagnostics becoming part of an integrated hybrid monitoring system combining physiological, metabolic, and behavioural data for personalised care,” she said. 

The potential 

When it comes to diagnostics, sweat has a lot going for it. It’s easier to collect than other samples like blood or urine, while allowing for real-time tracking of metabolites, electrolytes, and biomarkers.

As Dr Bordin and colleagues wrote in their review of the space, research teams are already focused on proving its potential in a number of different patient settings.

UTS lecturer in analytical chemistry, Dr Dayanne Bordin, co-authored the review of the diagnostic landscape for sweat.

Fields include endocrinology, where several studies have developed wearable sensors which use sweat samples for insights into glucose levels [link here].

Meanwhile, a study of patients with Alzheimer’s disease suggested those with the condition showed notably higher mean sweat Na+ concentration compared with a control group, and had impaired sweating responses to stimulation [link here].

Projects are also evaluating sweat composition for potential diagnostic markers for Parkinson’s disease, with one study detecting metabolites that could be associated with Parkinson’s phenotypes in sebum samples [link here].

Metals analysis through sweat and the study of volatile organic compounds (VOCs) emanating from the skin are also emerging applications of sweat testing.

Dr Bordin said future use cases included early disease detection.

“Sweat proteomics and metabolomics are revealing biomarkers linked to cystic fibrosis, diabetes, tuberculosis, Parkinson’s, and Alzheimer’s disease, allowing for earlier diagnosis and disease progression monitoring,” she said. 

The hurdles  

Despite its potential, there will be challenges on the path to sweat becoming a routine clinical tool.

“One of the most critical challenges before developing sensors for clinical use and point-of-care health monitoring is identifying clinically significant biomarkers and translating them into measurable analytes in sweat,” the authors explained.

Advanced analytical techniques have expanded the ability to identify complex metabolites and biomarkers in the lab.

While consumer-level wearables already exist on the market for tracking hydration and electrolyte balance, validating tests for medical-grade use is a steeper hill to climb.

“Broader studies are needed to correlate sweat biomarkers with blood equivalents, address inter-individual variability, and standardise sampling and normalisation,” Dr Bordin said. 

The ability for sweat tests to predict pathologies before they actually appear is a big challenge to overcome, the authors emphasised.

In order to become predictive, wearable devices need to not just record analyte concentrations in sweat, but to assess normal variations within an individual to work out what is noteworthy.

This is where AI has a role to play. “AI-driven data interpretation can enhance diagnostic precision when sweat data are integrated with other biofluids, enabling predictive and preventive medicine,” Dr Bordin said. 

Researchers are already showing how AI algorithms can help interpret sweat data. For example, one study published earlier this year showed machine learning models could interpret sweat pH and glucose levels through a stretchable, adhesive bilayer hydrogel-based patch [link here].

Dr Bordin and colleagues said demand for non-invasive testing meant sweat would stay in the spotlight for some time to come.

“However, the key challenge is moving from labour-intensive manual, endpoint-analyses, into a wearable, continuous monitoring system,” they wrote.

“This system requires interfacing of microfluidics, sensing, and computation within an electronic platform that is attached to a region of the body and that offers wireless connectivity to a personal device, from which an app will automatically be able to transfer data into meaningful information for the wearer.”