Can Big Data Improve Medical Diagnoses?
Mo Data stashed this in Big Data in Healthcare
That’s the idea behind Enlitic, a new startup founded by Jeremy Howard. Howard is a heavy hitter in big data: He’s data strategist for Silicon Valley venture capital powerhouse Khosla Ventures and former president and chief scientist of Kaggle, which hosts big-brain predictive modeling competitions.
Enlitic aims to use advances in machine learning to make medical diagnoses. The company says it has partnered with hospitals and medical imaging companies – Howard won’t say which ones – to mine a host of data sources, including x-rays, lab results, handwritten doctors notes, and claims records. The company is building software algorithms that identify otherwise invisible patterns in these data sources, making for sharper diagnoses.
The software, which Howard began developing a year ago, could help doctors diagnose illnesses more quickly and accurately, he says. In essence, it would concentrate the experience of thousands of doctors and patients. It would also be helpful in developing countries and rural areas where doctors are scarce. Less-trained clinicians would benefit from expert opinions encoded in software.
Of course, automating medical care is a risky business. It’s not hard to envision a faulty algorithm leading to an incorrect diagnosis. Howard insists that he isn’t attempting to replace physicians or enable them to automate medical care. “This should just be extra stuff on top of what they are already using,” he says.
Enlitic is part of a trend in which software engineers are moving into healthcare. Flatiron Health, a New York startup that recently raised $130 million dollars from investors including Google Ventures GOOGL +0.02%, was founded by two Google ad-tech engineers. Flatiron is just one of a number of big-data startups aiming to detect useful patterns in medical records. Howard’s project is unique, however, in the breadth of data sources he aims to combine.
Howard, who announced Enlitic’s formation on Friday, declined to disclose how much money he has raised, how many partners the company has, how many images it has scanned, or whether its algorithms have produced successful results.