Husband, Dad, Investor, Company Builder, Philanthropist.

Roam Analytics: What You Need to Know about Big Data and AI in Healthcare

Big data analytics and artificial intelligence solutions have entered the healthcare industry in a big way in recent years. Digital health solutions like fitness wearables are now collecting data about patients.

Unfortunately, while medical professionals have access to a treasure trove of important patient information, they often have no way to organize or quantify it. Even with the increasing use of digital health records, medical professionals are unable to convert data, particularly qualitative data, into a useable format. This means they can’t yet use this information in meaningful ways.

That’s where companies like Roam Analytics come in. Roam has produced a technology platform designed to quantify language data taken from patient records. This helps healthcare professionals take full advantage of the information available.

Large amounts of clinical data are required in order to perform predictive modeling, gain an understanding of patient population trends, and use available data to make informed choices about patient treatment plans. However, most healthcare systems are not designed to properly facilitate effective data capture.

Though electronic medical records are now more widely used, most are not able to create data sets of enough depth or consistency to be truly useful. Roam Analytics helps unlock language-based data by offering technological solutions to quantify it effectively.

big data

Roam’s Backers and Advisors

Co-founded by Alex Turkeltaub, the co-founder and former CEO of Frontier Strategy Group, Roam Analytics is backed by a diverse team of advisors. Its leadership is made up of former CEOs, administrators, and college professors from a variety of different academic and professional backgrounds.

Current CEO Lauren Demeuse was previously chief of staff to the CIO of Palantir Technologies. There, she led investments in machine learning and artificial intelligence technologies. Chief scientist Chris Potts also works as a professor of linguistics and of computer science at Stanford University, where his research focuses on developing computational models of dialogue, linguistic reasoning, and emotional expression.

Roam Analytics is also backed by a number of influential investors, including Sway Ventures.

Roam Unlocks Language Data

By leveraging machine learning and artificial intelligence solutions, Roam Analytics can take language data from disparate record-keeping systems and quantify it for use by healthcare companies. This creates a more complete picture of patient health and of their providers.

Language data can be quantified to gather data on patient mental states, obstacles to care, social and behavioral determinants of health, detailed symptoms and conditions, and diagnoses, both those confirmed and those ruled out.

By using this type of data, Roam can provide pharmaceutical companies or medical device companies with key insights. The technology has the potential to accelerate FDA approval and help sales and marketing teams to find doctors or treatment centers in their practice areas.

Roam’s Current Efforts in Healthcare

In 2018, Roam Analytics collaborated with Viiv Healthcare. They explored models to predict HIV health outcomes and presented their results and methods at the Aids2018 Conference. For the project, Roam obtained a combination of structured and unstructured electronic health record (EHR) data. The information was collected from sources throughout the entire United States.

The structured data consisted of basic elements like demographic variables along with other information like labs, diagnoses, prescriptions, and other information. Roam also used a large collection of anonymized notes from providers.

stethoscope

These also contained a significant amount of language-based information, some of it structured into phrases or excerpts. Roam’s technology then preprocessed many of these phrases in order to help create more meaningful “term features” to use in its models.

These term features were repurposed to create a Vector Space Model. This was intended to help standardize the terms and give more context to certain predictive phrases. The end goal of all this data collection and analysis was to help better understand the relationship between certain clinical phrases and positive or negative HIV outcomes.

In this way, healthcare professionals can look for relationships between specific symptoms or attempted treatments and a positive or negative outcome with HIV. This information has the potential to significantly help researchers and doctors looking for better treatments for the disease.

The technology has applications for many other diseases and treatment methods and modalities. Machines can track data in a way that human beings simply cannot replicate. The research implications are compelling.

Future Goals

The use of machine learning and artificial intelligence in healthcare has the potential to change the pharmaceutical industry in fundamental ways. No longer would one-size-fits-all drugs be the norm. Instead, pharmaceutical companies could create targeted drugs that would only benefit a small portion of the populations.

By using AI and machine learning, these companies would be able to prove that a drug will work for those few people it targets. This would enable pharmaceutical companies to provide only the drug that will have real benefits. It’s just one of many ways that big data analytics can potentially change the business of healthcare. Companies like Roam Analytics are hoping to do exactly that.