Welcome to Digital Health Briefing, a new email providing the latest news, data, and insight on how digital technology is disrupting the healthcare ecosystem, produced by BI Intelligence.
Have feedback? We’d like to hear from you. Write me at:
AI STARTUP RAISES $US25 MILLION, SIGNS IP DEAL WITH MEMORIAL SLOAN KETTERING: Paige.AI, a healthcare startup that applies artificial intelligence (AI) to cancer pathology, has come out of stealth mode and announced an exclusive deal with Memorial Sloan Kettering Cancer Center (MSK). The partnership gives Paige.AI access to MSK’s computational pathology intellectual property and its 25 million pathology slides – one of the largest repositories of de-identified pathology data in the world, which Paige.AI will use to feed its AI service. The startup also announced that it raised $US25 million in funding, led by Breyer Capital, whose CEO sits on the board.
Paige.AI and MSK hope that AI can help speed up the process of diagnosing some cancers, freeing up doctors’ time. Pathologists review dozens of slides to determine whether a patient has cancer, with only a few of these slides being relevant to the diagnosis. The company aims to have its AI do an initial review of the slides so doctors only need to review relevant slides to make a diagnosis. For now, MSK is Paige.AI’s only customer, but the company plans to expand its base over time.
Paige.AI joins a slew of tech companies and research labs that are using their AI expertise to augment the treatment and diagnosis of cancer:
- Microsoft is applying cloud computing, machine learning and AI to help improve cancer treatment as a part of its Healthcare NeXT initiative, according to Futurism. Like IBM Watson, Microsoft is using its platform to parse tons of health data, including research, images, and DNA information to augment and inform researchers and doctors.
- John Hopkins University is investigating how machine learning algorithms can be deployed to detect multiple early-stage cancers through blood tests. This could radically improve the rates of early cancer diagnosis, which could improve treatment outcomes.
- PathAI is another startup applying AI to pathology. The company, which raised $US11 million in its Series A funding round in November 2017, counts Philips and Bristol-Myers Squibb as its customers, according to The Wall Street Journal.
Investment and research intoAIand machine learning will continue to grow in 2018 as the technologies become must-haves for healthcare providers.Throughout 2017, the power of AI technology was on display. That’s unlikely to slow down in 2018 as firms work to prove the effectiveness of their solutions, not only to justify the massive investments being made, but also as a way to rise to the top of what’s becoming an extremely competitive market. Healthcare AI VC deal volume and funding hit a five-year high in 2016, with almost $US800 million in investments across 90 deals, according to TM Capital. If these technologies are successful in improving care, reducing costs, and boosting patient engagement, it’s probable we’ll see them become part of a new standard of care for patients.
Enjoy reading this briefing? Sign up and receive Digital Health Briefing to your inbox.
EMBRACE BECOMES FIRST FDA-APPROVED SMARTWATCH FOR USE IN NEUROLOGY: Empatica, a medical wearable developer, has become the first company to have its smartwatch cleared by the FDA for use in neurology. The smartwatch, dubbed “Embrace,” uses machine learning to monitor users’ seizures, and alerts caregivers when they become dangerous. The technology has proved to be quite effective in studies – Embrace’s algorithm detected 100% of the seizures in a multi-site clinical study of 135 patients diagnosed with epilepsy. Providing patients and physicians with an accurate way to track and report seizures could lead to advancements in how care is provided for those who suffer from conditions, such as epilepsy – 3.4 million people in the US suffer from the condition, according to a CDC estimate. The Embrace smartwatch has also highlighted how wearables can provide more accurate data for clinical studies. In the past, patients in clinical trials have had to self-report when seizures occur, taking a note in a diary. However, self-reporting has not been an effective way to track seizures – for example, over 40% of generalized tonic-clonic seizures are not reported by patients.
APPLE HEALTH RESEARCHER DEPARTS COMPANY: Stephen Friend, a prominent health researcher at Apple, has left the company after less than two years, according to CNBC. Friend was reportedly central to the development of Apple’s healthcare frameworks, HealthKit, ResearchKit, and CareKit – frameworks designed to help researchers, developers, and users access and record health data. In particular, Friend lent his expertise in security, privacy, and consent, helping Apple to navigate the highly-regulated healthcare system. It’s unclear how Friend’s departure, which CNBC says occurred in November 2017, will impact Apple’s future health offerings. Nevertheless, the company is likely to continue its forward momentum in the healthcare sector, given that Apple CEO Tim Cook has noted that healthcare is of massive interest to the company. Apple recently launched two big efforts in digital health, including the Apple Heart Study – in conjunction with Stanford Medicine and telehealth company American Well – and a beta update to the Apple Health app, which allows iPhone users to store and share their electronic medical records with certain US providers.
MEDIAL EARLYSIGN SHOWS AI AND EHR DATA CAN BE USED FOR EARLY DETECTION: Researchers from Medial EarlySign, a provider of machine-learning solutions, found that the combination of machine learning technology and electronic health record (EHR) data can be more effective than current clinical tools in identifying the risk of kidney damage in diabetics. The machine-learning-based model by Medial EarlySign used data found in EHRs, such as laboratory tests results, demographics, medication, and diagnostic codes, to predict a patient’s risk of experiencing renal dysfunction. The algorithm was able to identify nearly half of the patients who were expected to have significant kidney damage within a year. This figure represents 25% more patients than would have been identified by commonly used clinical tools. The algorithm, which is likely to be adopted by hospitals, insurers, and pharmaceutical firms, could become a go-to resource for reducing the likelihood of advanced renal disease. In the US, treatment of chronic kidney disease is expected to exceed $US48 billion per year, according to World Kidney Day.