Artificial intelligence (AI) is a technology that takes over when the human brain has reached its limit. For medical technology companies, it could help in the creation of more precise tools and devices that support diagnostic, therapeutic and surgical decisions. Instead of shooting an arrow into the sky, the bullseye target is clear, backed up by volumes of data and a machine learning algorithm designed to pinpoint the biomarker, the abnormality and the probability of success.
An analysis of deals tracked by Cortellis Deals Intelligence (CDI) and covered by BioWorld MedTech found a rising number of AI deals involving companies focused on medical device or diagnostic products. Such med-tech deals have climbed from only four in 2015 to a high of 32 last year. Already in 2019, there are 25, with five months to go. (See Volume of med-tech AI deals, right.)
This volume increase is mirrored by the biopharma industry and by an analysis of all digital health software deals within the CDI database.
AI deal values are typically not disclosed, but some stand out, such as a $125 million March partnership between Osaka, Japan-based Shionogi & Co. Ltd. and Boston-based Akili Interactive Labs Inc. to develop AKL-T01 and AKL-T02 digital medicines against attention deficit hyperactivity disorder and autism spectrum disorder in Japan and Taiwan. The treatments are delivered through immersive action video games.
"We have so many problems scientists are trying to solve," said Joel Haspel, vice president of digital health, life sciences, at Clarivate Analytics. "We have a massive amount of growing information. And over the last 10 years, the young people who were getting schooled in this stuff started applying the computer programming through the problems." Clarivate just launched Cortellis Digital Health Intelligence, a data solution that covers deals, independent health app reviews, digital health news and the latest discovery, development and commercialization trends.
Through a former employer, Haspel worked with the University of Warwick in the U.K. where pathologists and computer scientists applied AI technologies to auto-stage and grade colorectal cancer by having the machine count the cancer cells on the image. Warwick received £2.3 million (US$2.8 million) in funding last November from U.K. Research and Innovation for the technology.
Other institutions have begun similar experiments leading to academic spinouts. At the same time, the phrase 'outcome-based reimbursement' began circulating, motivating biopharma and med-tech firms to find better ways to discover and develop their products.
The possibilities are endless, Haspel said. By monitoring a senior citizen's gait, AI technologies can predict and thereby prevent a fall. For diabetics, it can monitor blood sugar levels and remind people to take their medications. If the technology could identify within a person's blood a marker of cancer before any tumors form, then an intervention may be able to prevent cancer entirely.
"Now, we're lucky if you catch it in stage one or two, and your life expectancy goes down," Haspel said.
In May, Darmstadt, Germany-based Merck KGaA, through its M Ventures arm, participated in the $6.3 million series A round of Houston-based Altoida Inc. for its Neuro-Motor Index AD medical device and brain health data platform for Alzheimer's disease. The product analyzes visuospatial and executive function during daily living activities through a battery of tests that ask patients to hide and find virtual objects in a physical space.
A number of other deals that employ AI technologies in 2019 involve sensors or implants to monitor symptoms or tumors and to adjust drug dosages. Other applications include improved diagnostic and predictive response methods. For instance, Celsius Therapeutics, of Cambridge, Mass., is identifying biomarkers in Horsham, Pa.-based Janssen Biotech Inc.'s Vega study of Tremfya (guselkumab) and Simponi (golimumab) for ulcerative colitis.
Janssen also is using AI for its Alzheimer's disease programs.
"We don't really have a good biomarker, a surrogate for efficacy. It just takes a long time and hundreds of millions of dollars to test a hypothesis today in Alzheimer's disease," said Vaibhav Narayan, the company's vice president and head of data science and digital health solutions for neurosciences, who spoke at the Biotechnology Innovation Organization's annual convention (BIO 2019) in Philadelphia in June. "Given the flexibility the regulators are showing, if we could find a biomarker that really tells us more quickly whether a drug is working or whether an intervention is working or not, that would be high on my wish list."
Overcoming hype, gaining knowledge
Philadelphia-based Nia Therapeutics Inc. in July acquired brain sensing and stimulation technology from Berkeley, Calif.-based Cortera Neurotechnologies Inc. to help create AI-driven brain implants to improve memory in traumatic brain injury patients. Parallax Health Sciences Inc., of Santa Monica, Calif., signed a licensing agreement for Austin-based Healthpoint Solutions, which is leveraging cognitive AI in health and wellness products. And Dublin-based Medtronic plc teamed up with San Francisco-based Viz.ai to speed adoption of its technology using AI to identify suspected large vessel occlusion strokes and automatically notify specialists. (See BioWorld MedTech, July 22, 2019.)
Several academic and government institutions also are involved in advancing AI for the medical technology industry. In July, Eagan, Minn.-based Predictive Oncology Inc., a subsidiary of Helomics Corp., signed on with the University of Pittsburgh Medical Center to study the use of AI to improve clinical decision making for ovarian cancer patients. They will use Helomics' D-Chip platform to analyze the genomic and drug response profiles of patients to determine their responses to therapy.
The University of Chicago paired with Chicago-based Paragon Biosciences LLC to commercialize its technology, Quantx, which is powered by machine learning AI and trained by a large reference database of abnormalities. It is cleared by the FDA to diagnose breast cancer. (See BioWorld MedTech, July 18, 2019.)
Cambridge, Mass.-based GNS Healthcare is one of about 150 AI companies focused on drug and biomarker discovery and other health care applications. The machine learning company completed a $23 million series D venture capital (VC) round in July.
"There are new companies all saying that they're AI and getting big VC money. There's a lot of hype there. But within the confines of an industry that's still under pressure to treat patients I think the buzz has led to people really trying to figure out how to do this," said Iya Khalil, chief commercial officer and founder at GNS, during BIO 2019. "We need to do it if we want to get to better treatments, but we have to then put in the hard work in figuring out who are the cohorts, who are the patients, how to get the right datasets, how to turn that into a learning system, as opposed to just a once-and-done and we throw away all of the data once we get our answer."
Human and machines working together
The Medical Futurist has tracked FDA approvals of AI-based algorithms in medicine, indicating an increase from one approval in 2014 to 23 in 2018, mostly for med-tech products. Through May of this year, there have been a handful of approvals, most recently for Mountain View, Calif.-based Alivecor's Kardiamobile, a smartphone electrocardiogram (ECG) device to detect heart arrhythmias; Kibbutz Shefayim, Israel-based Zebra Medical Vision's Healthpnx, an AI alert for pneumothorax based on chest X-rays; Tel Aviv, Israel-based Aidoc's pulmonary embolism diagnostic; and South San Francisco-based Verily Life Sciences LLC's ECG feature on the Verily Study Watch.
"You can just see the numbers increasing over the years. It's just continuing to grow," said Rick Finch, Clarivate Analytics' global head of consulting, life sciences.
Notably, New York-based Paige.AI, which was founded in early 2018, received breakthrough device designation from the FDA in March for the use of its AI technology in cancer diagnosis. The company's work grew out of efforts by Memorial Sloan Kettering Cancer scientists who have digitized more than 1 million pathology slides.
FDA guidelines of how to approve digital health technologies has likely contributed to the increase in investor interest and partnerships. While larger companies have bigger budgets, Finch predicts more mid-sized companies will begin to "latch on" as personnel develop more expertise and concerns are alleviated.
"The stuff that we're seeing rolled out now is more a combination of human and machine working together," Finch said. "It's not so much a job displacer as an empowerer of more impactful decisions."
Likewise, Dave Meyers, the National Director Life Sciences for Microsoft U.S. Health & Life Sciences, said during a presentation at the June Drug Information Association meeting in San Diego, that AI lacks the ability to mimic unique human qualities, such as reasoning, critical thinking and empathy. In other words, humans are needed.
"Artificial Intelligence is not about quantifying the obvious," Meyers said. "It's about opening our minds and processes to discover new things we have yet to consider."
Microsoft is working on Project Hanover with the Knight Cancer Institute at Oregon Health and Science University in a machine learning approach to personalize drug combinations for acute myeloid leukemia.
Other deals to note this year include Dublin-based Cosmo Pharmaceuticals NV partnership in April with Dublin-based Medtronic plc to distribute AI software and devices for the detection of lesions during colonoscopies. Oxford, U.K.-based Sensyne Health plc has signed deals to gather patient data from U.K. hospital foundations. In April, Leverkusen, Germany-based Bayer partnered with Sensyne and other consortium members to develop proposals to scale the network with hopes it will lead to new medicines, medical devices and biomarkers.
Sensyne and Bayer paired up again at the end of July to develop treatments for cardiovascular disease.
A number of other partnerships and venture capital investments are propelling the technology forward, and dealmaking is showing no signs of slowing, although the possibilities are not yet fully grasped.
"There's a lot of noise being created out there, and we haven't started separating the signal from the noise," Haspel said. "We haven't really seen the success. We're not there yet. We're still experimenting, developing, proving."