Making biomedical data computable through augmented intelligence
By Ikyatha Yerasala
nference, a Cambridge-based company founded by Indian scientists Venky Soundararajan and Murali Aravamudan, is making biomedical data computable to solve real life complex medical problems. Ikyatha Yerasala delves into this AI marvel.
Last year, when the Corona virus hit its peak in China and created absolute pandemonium, scientists in the country identified one of the two receptors by which the SARS-CoV-2, enters cells. Called the ACE2 receptor, it was the same gateway that the original SARS virus used. ACE2 intrigued scientists across the globe who delved into studying the receptor which would lead to remarkable discoveries and one of the significant ones would be the study done by nference, whose team decided to do a deep profiling of the ACE2 receptor using their company’s nferx software. The discoveries made by nference about how COVID-19 affects the gut, respiratory system and how the virus brings about anosmia (loss of smell) were path-breaking.
How did they do it? By using the nferX software, a cloud-based platform that enables users to dynamically query the universe of possible conceptual associations from over 100 million biomedical documents, including the COVID-19 Open Research Dataset announced by the White House. The team at nference combed these documents from published papers to genomic and other -omics databases, and uncovered multiple tissues and cell types with ACE2 receptors, and made a detailed report about it. “nferX is a powerful software platform. It synthesizes biomedical knowledge in a very comprehensive way. The idea is to replace a manual time-consuming and inefficient process, to produce real-world evidence in real-time by rapidly converting all these masses of data into insights that can advance the discovery and development of diagnostics and therapeutics,” shares Anita Kawatra, Chief Corporate Affairs Officer, nference.
What is nference?
nference, headquartered in Cambridge, MA is a company that’s using augmented intelligence to transform healthcare by making biomedical knowledge computable. Founded by entrepreneur, scientist and inventor Venky Soundararajan, and serial entrepreneur, scientist and engineer Murali Aravamudan, the company makes biomedical data accessible to solve real-world life sciences problems in real time. “We see ourselves as a science first software company that uses advanced deep learning and augmented intelligence to transform healthcare by making biomedical knowledge computable. We have a strategic exclusive partnership with Mayo Clinic, which is one of the world's top academic medical centers, to synthesize more than 100 years of institutional knowledge. Because the world has seen such a massive digitization of healthcare information, from scientific literature on molecular data, real-world evidence, coupled with advances in next generation sequencing, we have large scale initiatives. There’s siloed information - from the UK Biobank with 500,000 patients to clinical genomic data and also medical health records, which have created an incredible amount of real-world data about human health. There are also adverse event reporting systems to intellectual property records, SEC filings about what companies do - all this data creates incredible opportunity to solve complex medical problems. The data has to be made computable and then synthesized and curated in order to really solve problems and im-prove the lives of people on a global scale. Even with teams of experts, it would literally take decades to digest all this genomic data, molecular data and clinical and regulatory filings. And that's just what's in the public domain,” explains Anita.
Strategic partnerships
Nference ties up strategically with various organisations to use their nferX software to solve chal-lenges globally. Their collaboration with the Mayo Clinic is a significant one. “We are doing pio-neering work with de-identification, and the deidentification system is created by Indian tech-nologists. It adds electronic health records that have been deidentified using the system, labora-tory records, pathology, images, physician notes, other data representing the collective wisdom of thousands of physicians and millions of patients. This is rich data because it talks about biological context, therapeutic outcomes etc. We apply our algorithms to this data, and that can translate into earlier and better diagnosis, treatments, care, therapeutics and cures,” states Kawatra.
With about 150 team members in India, nference is looking to expand extensively this year. The company also has offices in Rochester, Minnesota, as part of their collaboration with the Mayo Clinic as well as in Toronto, Canada. They will continue to look at more strategic collaborators and partners.
COVID-related studies
nference’s scientists are part of the section of medical experts globally that’s fervently studying the SARS-Cov2 and discovering intriguing information about how it functions, enters the human system and the areas it affects human beings in. The Washington Post, which recently called nference the ‘Google of Biomedicine’, reported how AJ Venkatakrishnan, Director of Scientific Research and Partnerships at nference, made an accidental discovery about the new coronavirus, when he took genetic data from 10,967 samples and fed it into a machine. The scientist discov-ered that the virus had the DNA code — “RRARSVAS” — distinct from its predecessor coro-naviruses. With this sequence mimicking a protein that helps the human body regulate salt and fluid balance, it started various discussions about the virus. “We applied nference technology working with our own internal technologists and biomedical scientists, as well as all the clinicians at the Mayo Clinic to aid understanding of the SARS-COV2 virus – its transmission, symptoms and progression. Our platform really enabled some major discoveries from a molecular and clini-cal perspective. This work is really advancing the development, because it's advancing the way organizations can develop therapies and vaccines to curtail the burden of the virus - from your childhood immunizations to your ordinary immunizations that you would have over the course of your lifetime – we discovered that those have some level of protective benefit.”
A recent report reveals that an nference COVID-19 study and online tracker is the first to pro-duce evidence of increased COVID-19 vaccinations on decreased viral incidence by US coun-ties. Another interesting discovery made by nference about COVID treatment in the West, which resulted in discussions about how patients are treated is when they discovered racial disparities in anticoagulant treatment. “Through our data, we've seen that white patients were given a different version of an anticoagulant that actually is more effective than black patients. Our data showed that different versions of blood thinners heparin were given to white patients and black patients overall. The one that was more effective was given more often to white patients, and the one that was less effective to black patients. These discoveries are important to bring up because a lot of times they're reflecting unconscious bias. We noticed that the drug that was given to white pa-tients was actually more expensive. It's important that all patients have access to the treatment, that actually works better. We can identify what treatments work better, but we can also identify things like racial disparity in care that is worthy of examination,” reveals Kawatra.
The team at nference was also one of the earliest to identify loss of sense of taste and smell brought on by the coronavirus, which have now led to tests like the sniff test because taking one’s temperature isn't really an indicator of whether or not you have the virus. “These are ways we've enabled biopharmaceutical research to advance academic and scientific research by con-tributing to this understanding of every aspect of the virus. All of these enable both researches and clinical care to advance so that doctors and those on the frontline can use this information to advance care and scientists can use information as well, in the long-term,” she adds.
Discoveries about other illnesses
The company’s research on a life-threatening chronic condition - pulmonary hypertension, which is hard to diagnose, is commendable. “The average patient waits four to six years, from the first presentation of the symptoms to diagnosis because the symptoms are often mistaken for the symptoms of other diseases that are more common like asthma. The diagnosis itself is done by a very invasive, costly procedure. It's called right heart catheterization. We can predict the onset of hypertension in patients on an average of 18 months ahead of diagnosis, with more than 80% ac-curacy, and five years ahead of diagnosis with 75% accuracy. There are numerous other diseases that we will be able to apply our technology and algorithms to - various kinds of cardiac disease, different kinds of amyloidosis, heart failure and atrial fibrillation. We hope this will benefit a huge number of patients in the long-term.”
Patient privacy
When it comes to data analytics and third-party partnerships, there’s always the concern about patient privacy. How does nference ensure the protection of patients’ personal data? “We're centered around protecting patient privacy. We didn't create a solution and tweak it to protect patient privacy. We built the absolute best in protecting patient privacy and then built the system around that. We’ve established a secure privacy protecting environment, that is empowering researchers and clinicians to derive important insights that can accelerate research, the development of diagnostics, therapeutics and accelerate cures. Our goal is to improve patient outcomes, unlock these insights and biomedical data, and also protect patient privacy. It's not based on one person's data, it's based on the collective wisdom and experience of millions of scientists, physicians and patients. As for our collaboration with Mayo, that data never leaves the Mayo cloud. It's never sold to anyone, shared with anyone or copied. Researchers can only go in there, apply their output, apply the nference algorithms or apply their own algorithms. It's called a federated architecture where the data never leaves the cloud. The only thing you remove from the cloud is meaningful and clinical insights that can advance treatments, diagnosis care, and never any data itself,” shares Kawatra, explaining nference’s Clinical Data Analytics Platform, which is their collaboration with The Mayo Clinic.
By strategically using AI to make biomedical data computable and tapping into electronic health records and unstructured data, nference is revolutionizing healthcare and their partnerships with biopharma companies will only enable them to solve more complex global problems in the future.