Article

AI: A health-care game changer is here

Artificial intelligence is helping us to take a major step forward to manage population health.

When the World Wide Web first became global in 1991, few people would have suspected the magnitude by which the Internet would one day revolutionize how people gather information.

Even fewer would have foreseen its utility in helping health-care providers diagnose illness — but that’s exactly what happened nearly 30 years later when Microsoft began analyzing the search histories of users who had been seeking information about pancreatic adenocarcinoma using the company’s search engine, Bing.1 Not only did researchers discover that people diagnosed with pancreatic cancer initially begin searching for disease symptoms an average of 109.34 days prior to receiving their diagnoses, but, by analyzing their search histories, they found they could diagnose between 5% and 15% of adenocarcinoma cases with high fidelity (extremely low false-positive rates of 0.00001 and 0.0001) — no small feat for the fourth leading cause of cancer in the United States.1

“Five or 10 years ago, you’d see new treatment algorithms proposed for cancer every two years. Now that cycle of information is happening monthly, so oncology has changed dramatically as a result,” says John A. Hovanesian, MD, an ophthalmologist at Harvard Eye Associates in Orange County, Calif.

Microsoft’s research offers just one example of how organizations can tap into the utility of artificial intelligence (AI) to improve healthcare — a statistic that continues to rise: In 2017, 15% of all organizations used some form of AI, and that number is expected to keep growing, according to The Econsultancy 2018 Digital Trends, an annual report compiled by the software company Adobe.2

EXPECT TO SEE A LOT MORE OF IT

Despite the uptrend, AI is not a novel concept to the business world or the health-care industry. Some of the earliest documented uses of AI — devices that can perceive their environment and mimic “cognitive” functions of the other human minds, such as problem solving to maximize its chance of successfully achieving its goals — in medicine dates back to the 1970s when Stanford University researchers developed robotics that employed algorithms to select appropriate antibiotic therapy. The then-cutting-edge technology’s inferiority to infectious disease specialists halted its entry into the marketplace, but the instrumentation foreshadowed what was to come.3

Nearly five decades after AI made its medical debut, the health-care industry continues to benefit from its technological boons, such as improved efficiency, more accessible recordkeeping and enhanced compliance as the interweaving of health care and technology becomes increasingly more intimate.

Perhaps the recent announcement of the merger between distribution juggernaut Amazon and the online pharmacy PillPack is one of the latest major illustrations of this ongoing phenomenon. Licensed in all 50 states in addition to holding URAC and Verified Internet Pharmacy Practice Sites (VIPPS) regulatory certifications, PillPack’s technology simplifies medication regimens and improves adherence through what the company describes as “pre-sorted dose packaging, home delivery and customer service.”4

“AI will engender a new age of pharmacy benefit management,” says Dr. Hovanesian. “I think Amazon is looking to simplify the complex relationship between pharmacy benefit managers and pharmacy to change an industry that has become inordinarily complicated.”

He says the combination of PillPack’s unique business model and Amazon partnership may prove most beneficial in patients taking oral medications, as many ophthalmic medications come in multi-dose bottles without single-package or unit-dose features; however, he believes the ophthalmology community will still feel the ripples from the disruptive wave the partnership between Amazon and PillPack will create.

“With the power of Amazon, it’s been about the balance they bring to consumers. But, the data they’re able to collect from the ability to report outcomes and give patients those drugs will give us data we’ve never had before,” Dr. Hovanesian notes.

TAKE MEDICAL PRACTICE TO THE NEXT LEVEL

The growth of AI in health care has made its way to ophthalmology.

As an ophthalmologist with practices in two different states (Pennsylvania and New Jersey), Cynthia Matossian, MD, FACS, founder and director of Matossian Eye Associates, says that leveraging AI helps collate and analyze information to improve access to care and optimize patient outcomes as the health-care system continues to evolve. MatossianEye.com

In addition, ophthalmologists are finding that they can use AI to evaluate big data, allowing them to better manage their patient populations while simultaneously enhancing the quality and revenue of their practices. For example, Dr. Matossian relies on Conclusn, an analytics platform that uses AI to identify and target certain patients with disease-state-specific information about their condition, new medications and alternative treatment options — all based on ICD-10 codes in her practice’s EMR. She appreciates how the technology makes compiling and organizing information about a specific patient population less labor-intensive and more efficient than conventional paper-based documentation as well as an improvement on what she previously had in place in her EMR. (See “The key to optimize Big Data?,” October OM, page 96).

ELIMINATE WASTE, IMPROVE PATIENT SATISFACTION

Just as AI zeroes in on data trend patterns, it also eliminates waste by presenting patient information in a different way. Such is the case with glaucoma patients, who are typically seen by ophthalmologists every three to six months, a standard Dr. Hovanesian says is not medically necessary for some of these patients. AI can simplify big data related to such patients to identify patterns in disease, their causes (e.g., diseases linked to lifestyle, drugs, etc), and high-risk patients who require closer monitoring.

“Much of the way we treat disease is based upon dogma,” says Dr. Hovanesian. “We need to take the evidence AI provides and combine it with the dogma of medicine to improve care.”

Motivated by this realization and industry changes, he founded MDbackline, a company with a platform that optimizes big data analytics while strengthening patient education throughout all stages of cataract surgery. The platform also increases physician revenue by driving the conversion of cataract patients to premium intraocular lenses.

Another unique feature of MDbackline is that it encourages patients to provide feedback about their experience, which can help physicians improve their passive marketing.

“Most patients are happy with us, but most happy patients don’t get on Yelp and say so. Only the unhappy patients do, but this software will ask the patient to share testimonials [on websites],” says Dr. Hovanesian.

And success responses rates are high: Platform users enjoy a 70% response rate, which Dr. Hovanesian says gives physician practices previously unavailable data such as which implants patients really tolerate best, which patients might need to be brought in earlier for consideration of YAG capsulotomy and which patients might be willing to serve as a reference for future patients. The additional information allows physicians to continues the doctor-patient conversation to a level that office visit time constraints do not allow.

LOOKING AHEAD

Could AI hold the key to managing global population health?

Andrew Chang, head of Global Sales Devices for Zeiss and Meditec president of U.S. Sales and Service, believes AI will prove increasingly vital in addressing current and emerging issues in population health with the anticipated redistribution of the global population.

“The contraction and consolidation of eye-care practices in the global markets will continue,” Mr. Chang says. “With the growing population, patients needing care and shrinking number of providers, we have to find a more efficient way to deliver care in all countries.” He believes health-care systems face additional pressure to meet the demands of the burgeoning middle class growing in emerging markets, which he says will increase the need for additional care and delivery of that care.

AI can help ease the woes of health-care spending, which continues to soar along with our aging and growing population. Deloitte’s 2017 global life sciences outlook forecasts health-care spending to reach $8.7 trillion globally by 2020.5 Certain populations that typically require more health-care services are expected to contribute to that trend. According to Deloitte, the 65-and-over population will reach 656 million, or 11.5% of the total global, while the diabetic population is projected to increase to 642 million worldwide by 2040.6

“If you’re seeing 50 patients today but need to double the capacity tomorrow, health-care providers will need to change the way care is delivered,” Mr. Chang says. “The need for a more efficient and possibly a co-dependent network of delivering care is likely.”

“Currently, patients walking through our office doors encounter many hurdles — time management, the need to drive a car to get to our brick and mortar offices, taking time off from work, patients who do not drive need rides, etc.,” Dr. Matossian adds. One example of how to use technology to help patients with these hurdles is by placing non-mydriatic fundus cameras in primary-care practices or patient homes, she says. The ophthalmologist can read the images and make a diabetes diagnoses earlier in order to begin treatment sooner (see page 18).

FIND ANSWERS SOONER

Both Dr. Matossian and Mr. Chang believe remote screening and remote home care will become increasingly crucial in screening and diagnostics. Dr. Hovanesian believes that the shift in medical practice may be the most dynamic transition of all.

“Traditionally, research has been conducted by a curious investigator who saw a pattern in a disease and formed a hypothesis in which he or she suspected a relationship between a proposed cause and effect. AI eliminates the need for researchers to collect data refuting the hypothesis,” Dr. Hovanesian says. “Thanks to AI, we can use the data to find patterns we never thought of before instead of starting with assumptions.”

The story continues to unfold. OM

Disclosures: Dr. Hovanesian and Dr. Matossian are consultants for Zeiss.

REFERENCES

  1. Paparrizos J, White R, Horvitz E. Screening for pancreatic adenocarcinoma using signals from web search logs: feasibility study and results. J Oncol Practic. 2016;12:8:737-744.
  2. Abramovich G. Study finds investments in customer experience are paying off. CMO.com. Feb. 26, 2018. www.cmo.com/features/articles/2018/2/26/adobe-2018-digital-trends-report-findings.html#gs.aRlwymk%5D%5Bhttps://www.xaWXLSU .
  3. Schmidt-Erfurth U, Sadeghipour A, Genendas B, Waldstein S, Bogunovic. Artificial intelligence in retina. Retin Eye Res. 2018. [Epub ahead of print] www.sciencedirect.com/science/article/pii/S1350946218300119?via%3Dihub .
  4. Amazon to Acquire PillPack. Business Wire. June 28, 2018. www.businesswire.com/news/home/20180628005614/en/Amazon-Acquire-PillPack .
  5. Deloitte. 2017 Global health care sector outlook. www2.deloitte.com/content/dam/Deloitte/global/Documents/Life-Sciences-Health-Care/gx-lshc-2017-health-care-outlook-infographic.pdf
  6. Deloitte. 2018 Global health care outlook. The evolution of smart health care. www2.deloitte.com/content/dam/Deloitte/global/Images/infographics/gx-lshc-hc-outlook-2018-infographic.pdf