Combining proteomics and AI to enable ‘a new era in healthcare’

Alden Scientific’s AI-powered analysis tool measures individual risk for hundreds of health conditions, and the impact of potential interventions.

Understanding aging and age-related diseases requires analyzing a vast number of factors, including an individual’s genetics, immune system, epigenetics, environment and beyond. While AI has long been touted for its potential to shed light on these complexities of human biology and enable the next generation of healthcare, we’ve yet to see the emergence of tools that truly deliver on this promise.

Leveraging advanced plasma proteomics, US startup Alden Scientific has developed AI models capable of making the connections needed to accurately assess an individual’s state of health and risk of disease. The company’s tool measures more than 200 different conditions, including leading causes of morbidity and mortality such as Alzheimer’s, heart disease, diabetes and stroke. Significantly, its models also enable an individual to understand how an intervention impacts these risks.

With a host of top Silicon Valley investors among its early adopters, Alden is now using its platform to conduct an IRB-approved health study designed to provide a “longitudinal understanding of the interplay between environmental, biological, and medical data.”

Longevity.Technology: Alden was founded by serial entrepreneurs Jamie Heywood and Jeff Cole, the creators of health data platform PatientsLikeMe, which was sold to insurance giant UnitedHealth Group in 2019. While still operating in stealth mode, the company recently took the unprecedented decision to speak exclusively to LT about its work. We caught up with CEO Heywood, who explained why he believes Alden has cracked the code when it comes to the future of healthcare.

An MIT-trained mechanical engineer, Heywood turned to biomedicine after his brother was diagnosed with ALS in 1999, and ever since has been focused on building companies to improve the treatment and management of health and disease.

“From the ALS Therapy Development Institute to AOBiome and PatientsLikeMe, I’ve been involved in disease discovery, drug development and healthcare for 25 years,” he says. “And it’s given me a unique, personal view about what the medical and research system can and can’t explain, in terms of understanding how to intervene in an individual.”

Jamie Heywood, CEO, showing one of Alden’s models for atrial fibrillation. The company’s platform helps users understand their health state and future risk of disease, as well as progression to a disease state for themselves and a relevant population, and then model how behavioral changes or interventions alter the progression.

Better tools needed

Heywood and Cole realized that even with genetics and full healthcare data, modern medicine lacks the tools to understand human differences well enough to effectively guide individual care and definitively discriminate for whom, when and where a particular intervention might be effective.

“Whether it’s a stem cell transplant or a new drug for senescence or autophagy, or any new intervention, how do you know what will work for you?” says Heywood. “As an individual, is rapamycin going to help me, or Metformin, or sirtuins, or taking more fiber? What is going to change the course of my life? And the answer is, we don’t have the tools in care to tell us today.”

While he acknowledges that extensive work is going on around the world to try and find the answers to those questions, Heywood believes that nothing yet comes close to providing the depth of understanding needed to be effective in care. 

The tools and data are starting to emerge though. Today, biopharma and drug development AI companies are using advanced technologies to find new targets and understand health and disease. But these tools are expensive, hard to use and require rigorous quality control and standards. 

“Our idea is to use those same tools to guide individuals to their own best choices,” says Heywood. “Alden has built an end-to-end platform that enables that. Key to that is using these advanced research assays not just targeted validated clinical biomarkers and making that information useful in an individual.”

“People are making decisions about the most important thing for them, their health, with an extremely limited understanding of what actually drives their individual biology,” he adds. “What I’ve learned from having done early discovery and exploring fundamental new biology is that it did not have to be that way and that the technology to provide that understanding now exists; it’s just expensive and requires different skills to analyze and use.”

Leveraging LLMs for longevity

The tools that Heywood is referring to are based on machine learning mathematics, similar to the technology behind large language models, like ChatGPT combined with the most advanced multi-omic research bioassays like proteomics, metabolomics and other technologies that are used to discover novel biology.

“However, those tools have not yet been applied to look at the most advanced ways of measuring human differences – to establish for whom an intervention may make a difference on something they care about,” he says. “Whether that’s Alzheimer’s, heart disease or cancer risk, or grip strength, fluid intelligence, or even balding.”

Heywood says that Alden is aiming to refocus the power of those advanced AI tools on understanding health at an individual level.

“We are using those technologies today – we have built one of the world’s best machine learning and biology informatics teams, and we are connecting that data to find out what individual differences, timeframes, and contexts actually mean,” he says. “I believe we’re only at the very beginning of the revolution of digital medicine.”

Under the hood of Alden’s AI

Alden has assembled several large public and proprietary datasets that it uses to build its models. A core training data set for the company’s AI models comes from the UK Biobank’s recently released proteomics dataset of more than 50,000 individuals. Each record contains a vast amount of information, including genetics, medical records, labs, metabolomics, proteins, activity monitoring, questionnaires, repeat baseline assessments, blood and urine tests, imaging and more.

From a user perspective, the primary source of Alden’s data is an advanced blood test – leveraging a proteomics assay that measures 3,000 proteins in plasma, which Heywood says will soon rise to 5,500.

“We have built the most advanced AI models to understand how these proteins interact with an individual’s genetics, immune system, epigenetics, environment, infection, and even what they eat and breathe,” he says. “For example, compared to the current best risk estimates, we believe our environmentally responsive model provides approximately a 50-times more accurate assessment of an individual’s risk for developing Alzheimer’s over the next 10 years.”

Enabling a new era in healthcare

Crucially, in addition to assessing risk, Alden also aims to provide insight into what is driving these risks in an individual, and how that risk changes over time based on age and what actions they take.

“We’re able to create ‘what if’ scenarios that show the effect of an intervention on a particular risk and later record the changes to this risk based on the outcome of an intervention,” says Heywood.

Of course, the advanced testing and analysis that Alden is doing at this stage does not come cheap, which is why the first users of its tool will be “ultra-high-net-worth” individuals. A veritable Who’s Who of Silicon Valley investors has already signed up (we’ve seen the list), and these early users will form the first cohort in “Aurora Discovery” – an IRB-approved longitudinal health study designed to demonstrate the effectiveness of the technology.

“We’re at the beginning of a journey, and the first users of our tool are the adventurers, the people we’re working with to build a map of human health, just like the early oriented orienteers used GPS,” says Heywood. “However, we have a plan to drive scale and cost so that, over time, this can become a test that is available to all at a price point that reduces the cost of care while improving outcomes across the board.”

“Imagine having a cost-effective accurate digital representation of not just diseases but health itself. We believe Alden Scientific will ultimately become a crucial component of the 500 million primary care office visits each year. And I think that will enable true prevention and a new era in healthcare.”

Photographs courtesy of Alden Scientific