An international research team involving Helmholtz Munich (DZL site CPC-M) has developed an ultra-thin sensor patch that can simultaneously measure several physiological parameters. The silk-based system uses very small amounts of bodily fluid and combines colorimetric sensing with artificial intelligence. The study, published in ACS Sensors, was supported, among others, by the German Center for Lung Research (DZL).
Medical monitoring of premature infants requires regular checks of important health parameters, including temperature, glucose and sodium concentrations, and pH levels. Some of these measurements currently require blood sampling or other diagnostic procedures. A research team from Helmholtz Munich and international partner institutions has now developed a sensor system that can non-invasively measure several of these parameters.
Multiple measurements with a single patch
The patch, which is only a few millimeters thick, consists of several layers. The base layer is made of silk fibroin, which is derived from silkworm cocoons. A paper-based microfluidic layer absorbs tiny amounts of fluid and directs them to the respective sensor fields. In total, twelve different dyes respond to different biological parameters by changing color. The system can measure temperature, pH, and sodium and glucose concentrations. Potential sources of fluid include sweat, saliva, and interstitial fluid.
A key feature of the system is its AI-supported analysis. A deep-learning model analyzes the color changes while accounting for factors such as variations in lighting, camera angle, and movement. Another AI model can detect and track the patch through the plastic walls of an incubator.
Initial studies show promising results
In the studies conducted so far, the mean deviation for temperature measurements was approximately 0.5 °C, while pH measurements deviated by less than 0.5 pH units. Clinically relevant thresholds, including hypoglycemia and hypernatremia, were detected with an accuracy of more than 91 percent. For low glucose levels, the accuracy exceeded 98 percent.
The sensor system was also able to detect salt concentrations relevant to the diagnosis of cystic fibrosis. The patch takes advantage of the comparatively high fluid loss through the skin of premature infants. Within approximately 15 to 40 minutes, sufficient fluid can be collected to analyze the four parameters studied. Only a few microliters are required per measurement point.
The studies conducted to date have included saliva and sweat samples from adults. The sensor system was tested under various conditions, including fasting, physical exercise, and meals. The measured changes showed trends similar to those observed with established laboratory methods and finger-prick blood glucose measurements. “We do not aim to replace laboratory diagnostics,” says Benjamin Schubert, head of a DZL-funded research group at the Computational Health Center at Helmholtz Munich. “Our goal is to detect changes that occur between two laboratory tests.”
Next steps
The current study represents a proof of principle. Further research will be necessary before the technology can be used in neonatology. Larger studies in neonatal intensive care units are planned to compare the patch’s results with conventional blood analyses. The AI system will also be further developed using data from different incubator models, lighting conditions, and skin tones. Additional toxicological and regulatory studies are also planned.
In the long term, the technology could be expanded to include additional measurement parameters. Since the patch requires no power supply, wiring, or cooling, the researchers are also investigating other potential applications for this low-cost sensor technology.
The study was supported, among others, by the Helmholtz Association, the German Federal Ministry of Education and Research (BMBF), Helmholtz Munich, and the German Center for Lung Research (DZL). It is also part of the transregional Collaborative Research Centre PILOT (TRR 359, “Perinatal Development of Immune Cell Topology”), funded by the German Research Foundation (DFG).
Source: Helmholtz Munich and Partners Develop AI-Powered Sensor Patch for Neonatal Care
Original publication: Castelblanco A, Ruggeri E, Matzeu G, Heydarian M, Foerster K, Bahnasy A, Flemmer A, Schnabel JA, Schubert B, Omenetto FG, Hilgendorff A. Artificial Intelligence-Supported Colorimetric Multibiomarker Sensor to Enable Critical Neonatal Monitoring. ACS Sens. 2026 Jun 26;11(6):4409–4419. doi: 10.1021/acssensors.5c04171. Epub 2026 May 28. PMID: 42205010.
