Andrew Lacy

New biomarkers are changing the way we screen for illnesses

May 03, 2021
By Andrew Lacy

Thanks to recent breakthroughs in medical imaging, biochemical and genetic research, and AI-enabled analytics, a wave of new biomarkers are now changing the ways we think about a wide range of illnesses, from new challenges like COVID-19 to longstanding health concerns such as immune disorders and gastrointestinal ailments. The resultant flood of data can create challenges for physicians — but in the long run, novel biomarkers will give us powerful new clinical weapons as we fight for our patients' health.

In search of better treatments
In some cases, biomarkers can offer entirely new pathways to prevention or treatment. It was recently found that babies with high levels of heavy metals in their teeth are more likely to develop Crohn's disease, for instance — a remarkable environmental factor that could make it easier to encourage healthier behaviors in mothers or caregivers. Similarly, lupus sufferers can increasingly look forward to customized treatments as doctors begin to use novel biomarkers found in blood or urine to monitor their disease's progression and response to clinical interventions.

Of course, novel biomarkers are especially important when we're tackling novel diseases, and the COVID-19 pandemic has driven important new research in this area. Some researchers have identified specific proteins that can indicate an increased likelihood of serious disease, helping healthcare professionals to triage patients and identify suitable candidates for antibody therapies and other new interventions that can't be offered to every single infected patient.

The possibility of forecasting a patient's death with 90% accuracy up to 10 days in advance might sound like something lifted from the pages of a science fiction novel. Still, the reality is that overburdened clinicians are already having to make judgment calls of this sort on a daily basis. In a world where doctors already need to make difficult choices about which patients get oxygen, ventilators, antibody therapies, or other potentially scarce resources, we should at least ensure we're equipping them to make smart, well-informed decisions along the way.

The power of AI
Making sense of biomarkers isn't always as simple as running an assay and getting a straightforward, yes-or-no result. Often, especially in the world of medical imaging, human judgement and professional experience play a huge role in parsing out the clinically useful signals amidst the noise. In such areas, improved analysis of existing data can be as important as finding novel markers. Machine learning tools and neural networks are now able to distinguish COVID-19 infections from community-acquired pneumonia with about 90% accuracy based on CT scans, for instance, and can also help imaging specialists to diagnose cancers and other illnesses earlier in their progression.

In coming months and years, we'll see AI analytics used to unlock even richer sources of biomarkers. One MIT team has successfully developed an algorithm that can accurately distinguish COVID-19 infections from other illnesses based on the sound of a cough recorded via a cellphone call, for instance. Given the flood of new research being published as health systems around the world grapple with COVID-19, computerized meta-research is also proving vital as we sift through the data to uncover clinically salient biomarkers.

Advanced whole-body imaging, which looks not just at each organ but the body as a whole, is also opening the door to new clinical insights is also opening the door to new clinical insights, with researchers able to rely not just on structural changes in physical organs, but also on functional data showing blood flow, metabolic activity — and, yes, even the accumulation of "sweat" around certain organs. Used in conjunction with machine learning technologies, such approaches can also tease out complex network effects, such as identifying variances in the way that the brain, lungs and legs all interact in patients with cardiovascular disease. These interactions can serve as new biomarkers that can lead to even earlier warnings of an underlying medical condition. Even before symptoms become apparent to patients.

Too much information?
Are there downsides to all these biomarkers? Well, perhaps in some cases. Over-diagnosis of diseases that are relatively benign can be a real concern: we spot far more early-stage thyroid cancers than we used to, for instance, but because many thyroid tumors are extremely slow-growing, that means we're conducting more thyroidectomies (and shunting more patients into lifelong post-thyroid treatment protocols) without necessarily improving clinical outcomes.

In other cases — as with those COVID-19 prognostic biomarkers — more information can create ethical headaches. Nobody wants to deny patients life-saving treatment based on a blood test or an MRI scan. Still, as long as healthcare professionals are constrained by limited resources, from ICU beds to transplant organs, there will always be an aspect of triage and rationing to the work they do — and if we have to have rationing, it's better for it to be based on data and scientific insights.

The reality is that while healthcare professionals sometimes wind up having to make difficult decisions, there's no such thing as too much information in the medical sphere. Change is always unsettling, but with AI tools on hand to help us make sense of a flood of new data from imaging tools and assays, we're making smarter decisions and delivering better care to our patients. In months and years to come, novel biomarkers will empower us to diagnose illnesses earlier, deliver more targeted and effective care, and unlock powerful new treatments for diseases of all kinds. For both patients and healthcare professionals, that's unequivocally good news.

About the author: Andrew Lacy is the CEO of San Francisco-based Prenuvo.