In many cases, patients could receive a diagnosis and personalized treatment plan without leaving home. Hospital visits could be limited to treating the critically ill, performing surgery or diagnosing particularly complicated cases. Importantly, this would free up more doctors’ time for bedside care of those who critically need them.
Patient treatment planning will be based on an individual’s age, sex, race, general health, genetics, diet, allergies, lifestyle and more. It may even be possible to tailor drugs to individual needs and dosage requirements. The expectation is that individualized plans will bring increased precision to treatment, reduce the risk of overdosing and lead to safer and better outcomes.
Fewer interactions with healthcare professionals may seem antiseptic and lacking in humanity. However, while traditional house calls have already become as scarce as buggy whips, communications technology may enable our primary care professionals to check in with their patients at home throughout the day via a smartphone, mobile device or even through the television. Thus, while healthcare professionals may not be sitting at the patient’s side, they may be able to spend more time than ever consulting with those in their care.
AI and the Bottom Line
AI also offers the promise of reducing costs. Here are a few examples of what it could do:
- Bring greater efficiency and accountability to healthcare
- Reduce malpractice, clinical waste, administrative complexity and staff time spent on repetitive tasks
- Offer the precision of robot-assisted surgery, which can speed recovery and shorten hospital stays
- Use diagnostic imaging systems that identify illness earlier, resulting in shorter, less radical treatment plans that can save patients money
- Potentially assist young medical professionals with insights and recommendations
What’s Holding Back Progress?
In 1984, a robot at UBC Hospital in Vancouver, British Columbia, first assisted in an orthopedic surgical procedure. Since then, robot-assisted surgery has grown in popularity. Smaller, more precise cuts reduce risks of infection, speed recovery and shorten hospital stays. Yet it was still 22 years later when the first unassisted AI “doctor” performed surgery to correct a heart arrhythmia.
In 2013, researchers reported that after teaching IBM Watson the equivalent of a second-year med student, providing access to 1.5 million patient records and preparing it to process peer-reviewed medical research, the supercomputer had a 90% success rate in diagnosing lung cancer. The human doctor was successful just 50% of the time.
With demonstrable results, why is AI still primarily a promise for the future? Although there may be many reasons, including fear of change, here are three questions that come to mind:
1) What will it cost?
Hospitals, clinics and insurance groups will need to budget for significant capital expenditures for new equipment. Healthcare professionals will have to allocate time for specialized training. Also, computer programmers must work closely with healthcare professionals to design and test therapeutic software. However, if the benefits of shorter hospital stays, faster healing, earlier diagnosis and more proactive healthcare materialize as expected, they will help to offset the costs.
2) Is it safe?
Medicine often comes down to life or death decisions. Healthcare professionals are required to make dozens of judgments every day that can change patients’ lives—for the better but sometimes for worse. Are we ready to turn those decisions over to AI algorithms? While computers never tire and can process more information more quickly than the human brain, what happens when the program is in error? The answer is we will need to test every algorithm, every routine and each result rigorously. Even after testing is complete, it is likely that doctors will use machines in the role of an assistant rather than allowing them to rule supreme over medical decisions.
3) Is the data secure and patient privacy protected?
The only way deep learning will lead to new, faster, earlier diagnoses is with access to massive amounts of data from patients’ electronic health records. Here, the technology runs headlong into privacy and security imperatives. Since 1996, HIPAA (Health Insurance Portability & Accountability Act) requirements have put the full power of federal legislation on the side of the patient. Although these huge datasets may hold the key to quick, early diagnoses and new insights, they cannot be accessed and analyzed at the expense of individual privacy.
In all likelihood, technologies will continue to converge, AI algorithms and programming will become fully vetted, and healthcare facilities will begin to make the major capital investments necessary. Also, systems will be developed to ensure the protection of patient privacy. When all of this happens, healthcare of the future will become our reality. How closely the transition resembles the visions of medical futurists is yet to be known. What really matters is that AI can deliver on its promise and potential to augment medical insight, create a more personalized form of patient care and help control costs while ever improving outcomes. Above all, it can increase healthcare efficiencies, allowing doctors to dedicate more time to communicating with patients whether remotely or face to face.
[1] International Journal of Computer Theory and Engineering — Artificial Intelligence in Personalized Medicine Application of AI Algorithms in Solving Personalized Medicine Problems
[2] Economy & Markets — Diagnosed by a Machine: Artificial Intelligence Takes Over from Doctors
[3] Dr. Brian Day — World’s First Surgical Robot in B.C.
[4] Engadget, Evan Blass — Robot Surgeon Performs World’s First Unassisted Operation
[5] Wired — IBM’s Watson is Better at Diagnosing Cancer than Human Doctors
