New screening tool uses infrared and machine learning to detect multiple health conditions

Using infrared light and machine learning, researchers have developed an effective way to screen for human health conditions and deviations at a population level.

Imagine a scenario in which just a drop of blood could provide comprehensive information about your health within minutes. Thanks to recent scientific advances, this vision could become a reality. Scientists from the BIRD team led by Mihaela Žigman from the Ludwig Maximilian University Munich (LMU) and the Max Planck Institute of Quantum Optics (MPQ), in collaboration with the Helmholtz Center Munich, have developed a health screening tool that uses infrared light and machine learning to detect multiple health conditions in a single measurement.

Infrared spectroscopy, a technique that uses infrared light to analyze the molecular composition of a substance, has been a fundamental tool in chemistry for decades. It is like fingerprinting molecules with a specialized machine called a spectrometer. When applied to complex biological fluids such as blood plasma, this physicochemical technique reveals detailed information about molecular signals, making it a promising tool in medical diagnostics. Despite its long-standing use in chemistry and industry, infrared spectroscopy is neither established nor incorporated into the standard of medical diagnostics.

A team of scientists from the BIRD group at LMU and MPQ, led by Mihaela Zigmann, launched an effort to tackle this problem. They had previously established a method to measure human plasma and, working with Annette Peters’ team at Helmholtz Munich, pioneered infrared molecular fingerprinting for naturally diverse populations. This involved measuring the blood of thousands of individuals from the KORA study, a comprehensive health research project founded in Augsburg, Germany. Randomly selected adults, chosen as representative of a naturally varying population, were recruited for health checks and blood donations.

A wide range of potential applications

What is the value of the current research? The existing KORA study gained new value because it was tested from a new perspective and served a new purpose. More than 5,000 plasma samples were measured using Fourier transform infrared (FTIR) spectroscopy. Tarek Eissa and Cristina Leonardo from the BIRD team at LMU analyzed the blood samples from the KORA study using infrared light to obtain molecular fingerprints. The team applied machine learning to analyze the molecular fingerprints and correlate them with medical data. They found that these fingerprints contained valuable information that allowed for rapid health screening. Multitasking computer algorithms now make it possible to distinguish between different health conditions, such as abnormal levels of blood lipids, various changes in blood pressure, and the detection of not only type 2 diabetes but even prediabetes, a precursor to diabetes that often goes undetected.

Interestingly, the algorithm was also able to identify individuals who were or remained healthy throughout the study period. This was very important for two reasons. First, while most people in a randomly selected population will experience abnormal health changes, finding perfectly healthy individuals is hardly easy, as we are all different and change over time. Second, many people suffer from multiple symptoms in various combinations. Traditionally, doctors would need a new test for each disease. But this new approach doesn’t just identify one symptom at a time, it pinpoints a range of health issues. This machine learning-powered system not only identifies healthy individuals, but also detects complex conditions involving multiple diseases simultaneously. Moreover, it can predict the onset of metabolic syndrome years before symptoms appear, providing an opportunity for intervention.

The researchers say that this work lays the groundwork for infrared molecular fingerprinting to become a routine part of health checkups, allowing doctors to detect and manage medical conditions more efficiently. This is particularly important for metabolic disorders such as cholesterol abnormalities and diabetes, where timely and effective intervention could significantly improve outcomes. But the potential applications of this technology go much further. As the researchers continue to refine the system and expand its capabilities through technological development and establishment in the context of clinical studies, they expect that even more health conditions and combinations of them will be added to the diagnostic repertoire. This could lead to personalized health monitoring, where individuals check their health conditions regularly and catch potential problems long before they become serious.

In conclusion, the researchers believe that the combination of infrared spectroscopy and machine learning could revolutionize medical diagnostics. All it takes is a drop of blood and infrared light to create a powerful new tool that can monitor our health, spot problems more efficiently, and improve healthcare worldwide.