Breaking News
January 20, 2018 - MSU scientists seek to identify brain mechanisms related to psychosis
January 20, 2018 - Syndax Pharmaceuticals Announces Clinical Collaboration to Evaluate Entinostat in Combination with anti-PD-L1 Cancer Immunotherapy in Breast Cancer
January 20, 2018 - Endoscopes Over Microscopes in Retinal Surgery: Ophthalmology Times
January 20, 2018 - Technology not taking over children’s lives despite screen-time increase
January 20, 2018 - Study finds extensive contamination around lead battery recycling plants in 7 African countries
January 20, 2018 - Flu may pass to others through exhaled breath, study shows
January 20, 2018 - Neuronal loss very limited in Alzheimer’s disease, new study shows
January 20, 2018 - Novel robot can aid treatment of rare birth defect
January 19, 2018 - TherapeuticsMD Announces Submission of New Drug Application for TX-001HR
January 19, 2018 - Fighting Infant Mortality | Medpage Today
January 19, 2018 - Researchers offer new evidence on four-year-old children’s knowledge about ecology
January 19, 2018 - Analysis finds overlooked crucial factor in determining prognosis for DIPGs
January 19, 2018 - Review explores consequences of genetic testing and cancer risk-reducing surgery
January 19, 2018 - Morning Break: HHS Div. of Religious Freedom; Trump’s Heart Health; Minister of Loneliness
January 19, 2018 - Parkinson’s disease ‘jerking’ side effect detected by algorithm
January 19, 2018 - New analysis finds dramatic increases in maternal mortality rates
January 19, 2018 - Weight-Loss Surgery’s Benefits Wane Over Time for Diabetics
January 19, 2018 - Cath Lab Recap: Sapien 3 Delivery System Recall; Transatlantic PCI Smackdown
January 19, 2018 - Parkinson’s treatment could be more effective, student finds
January 19, 2018 - New vaccine approach offers effective protection against tuberculosis
January 19, 2018 - Home care agencies often wrongly deny Medicare help to the chronically ill
January 19, 2018 - One hundred percent fruit juice does not alter blood sugar levels
January 19, 2018 - Prebiotics could enhance learning and memory skills in infants
January 19, 2018 - CMS May Cover MRI With Cardiac Devices Across the Board
January 19, 2018 - As income rises, women get slimmer—but not men
January 19, 2018 - Researchers develop adhesive materials to prevent bracket stains on teeth
January 19, 2018 - Flu can be spread without coughs and sneezes
January 19, 2018 - AMSBIO’s new recombinant protein shows great promise for organoid culture
January 19, 2018 - AbbVie’s Upadacitinib Granted Breakthrough Therapy Designation from the U.S. Food and Drug Administration for Atopic Dermatitis
January 19, 2018 - ASH: Pomalidomide Dose Escalation Improves Response
January 19, 2018 - Is your child’s school an obesity risk?
January 19, 2018 - Scientists describe groundbreaking training effect on the innate immune system
January 19, 2018 - MAST announces new AmpC, ESBL & Carbapenemase Detection Set
January 19, 2018 - Signaling molecules likely involved in concussions, rodent studies show
January 19, 2018 - Mast introduces Carba plus for CPE and OXA-48 confirmation
January 19, 2018 - Paleolithic diet helps overweight women maintain weight loss
January 19, 2018 - Agios Submits New Drug Application to the FDA for Ivosidenib for the Treatment of Patients with Relapsed/Refractory AML and an IDH1 Mutation
January 19, 2018 - This Flu Season, Don’t Forget About Tamiflu
January 19, 2018 - Amsterdam wins battle to host EU medicines agency after Brexit
January 19, 2018 - Study suggests movement as accurate method to diagnose neurodevelopmental disorders
January 19, 2018 - Maximize resolution in deep imaging for neuroscience research with Olympus TruResolution objectives
January 19, 2018 - Bilingualism may benefit children with ASD
January 19, 2018 - FDA Alert: Levofloxacin in 5 Percent Dextrose 250mg/50mL by AuroMedics: Recall
January 19, 2018 - USPSTF Not Backing Ankle-Brachial Index, CRP, or Coronary Calcium
January 19, 2018 - Higher omega-3 fatty acid intake tied to lower glaucoma risk
January 19, 2018 - Findings reveal conventional cancer therapy as double-edged sword
January 19, 2018 - Health Highlights: Jan. 16, 2018
January 19, 2018 - Morning Break: Anti-Emetic Warning; Uninsured Rate Jumps; Flu Worsens Saline Shortage
January 19, 2018 - Increased use of ambulatory surgery centers for cataract surgery
January 19, 2018 - Not-for-profit hospitals coming up with their own generic medicines to combat shortages
January 19, 2018 - $500 cancer detection blood tests may soon become reality
January 19, 2018 - Chronic traumatic encephalopathy may start early even without signs of concussions
January 19, 2018 - Warm-up program for children cuts soccer injuries by 50%
January 19, 2018 - ‘You’re Old and You Need Tests’: What We Heard This Week
January 19, 2018 - Egg-preserving hysterectomy raises heart risks later: study
January 19, 2018 - GA-map Dysbiosis Test identifies IBS patients who respond to FODMAP diet, study shows
January 19, 2018 - Study explores mortality and health-related habits in former elite athletes and their brothers
January 19, 2018 - New biodegradable sensors could assist doctors
January 19, 2018 - Modular gene enhancers may be suitable target in treatment of blood cancer
January 18, 2018 - New precision medicine trial for metastatic pancreatic cancer
January 18, 2018 - Shire Receives FDA Breakthrough Therapy Designation for Maribavir, an Investigational Treatment for Cytomegalovirus (CMV) Infection in Transplant Patients
January 18, 2018 - Pre-Existing Patient-Valve Mismatch Trips Up ViV Implant
January 18, 2018 - Adolescents: health risks and solutions
January 18, 2018 - US woman delivers baby from embryo frozen for 24 years
January 18, 2018 - Study identifies new target for treatment of depression
January 18, 2018 - LJI study reveals key player that promotes skin inflammation in atopic dermatitis
January 18, 2018 - Study devises efficient and economical strategy to screen breast and ovarian cancer gene mutations
January 18, 2018 - Agile Therapeutics, Inc. Receives a Complete Response Letter from the FDA for Twirla (AG200-15) for the Prevention of Pregnancy
January 18, 2018 - Gene Therapy for Inherited Retinal Dystrophy Gets FDA Clearance
January 18, 2018 - Researchers identify a new chemical pathway that helps the brain detect sweet, savory and bitter flavors
January 18, 2018 - IBV develops platform that helps companies to diagnose wellbeing of their workforce
January 18, 2018 - Study to test new precision medicine approach for metastatic pancreatic cancer
January 18, 2018 - World’s first vaccine relieves grass pollen allergy symptoms by at least 25%, study shows
January 18, 2018 - FDA Approves New Indication for Gilotrif (afatinib) in EGFR Mutation-Positive NSCLC
January 18, 2018 - Oncologists Dish on Top Issues for 2018
January 18, 2018 - Researchers identify new potential drug target for Huntington’s disease
January 18, 2018 - Metrohm USA welcomes employees to new headquarters in Florida
January 18, 2018 - Human waste remains main source of fecal pollution in the river Danube
January 18, 2018 - Expert discusses how to stay healthy during flu season
January 18, 2018 - New biomaterials-based system improves T-cell production
Augmented Intelligence in Medicine | Medpage Today

Augmented Intelligence in Medicine | Medpage Today

image_pdfDownload PDFimage_print

It’s likely you have heard the news that radiologists, dermatologists, and the other visual specialists will soon have to look for new jobs. Machine learning using deep, convolutional, neural networks will automate diagnosis and replace these doctors (so the thinking goes). I have met more than a few armchair medical futurists excitedly making these claims. So, let me go out on a limb and say that just as I do not believe we will soon be colonizing Mars, I do not believe artificial intelligence (AI) will replace the need for doctors anytime soon.

I am not a Luddite; I disclose here that I am CEO of a company deeply engaged with machine learning (as well as a practicing dermatologist). AI will be transformative; I just do not believe that technology alone will engineer the US healthcare system out of the poor quality of care, mind-boggling waste, and suboptimal outcomes that define it. Let us never lose sight of the fact that of 17 industrialized nations, we consistently rank 16 or 17 on almost all basic health outcomes.

We should see all forms of information technology — including AI — as tools to help us be more accurate, gather more precise data, and see patterns across the population, and ultimately to develop new knowledge. Charles Friedman, PhD, pointed out years ago that the equation for information technology in health care is not “computer > doctor,” but rather “computer + doctor > doctor alone.”

Two years ago, I attended a national healthcare meeting where a cardiologist presented on the reduction in cardiac-related illness since the 1950s, illustrated with a slide that showed the reduction in cardiac disease burden and death over the past 50 years in the United States. The cardiologist proudly spoke to advances in cardiac imaging of atherosclerotic disease and coronary artery stents as the reason for success. As soon as he finished thumping his chest, with exquisite timing, a physician in the audience stood up and asked, “Doesn’t the timeline of your chart completely correlate with the reduced tobacco use in the US during this period?” With quite a bit of laughter and some applause, most of the physicians in the room were tacitly acknowledging that it was ridiculous to ascribe the progress we have made in reducing cardiac disease to the practice of medicine and not to the practice of public health. Tobacco use in the population went from close to 45% in the 1950s to about 18% during this interval.

Likewise, when thinking about how best to take advantage of AI, let’s make sure we are not getting ahead of ourselves regarding what technology can really do. Machine learning in imaging might be highly sensitive but not specific, thereby promoting false positives. Today’s diagnostic imaging, such as MRI, CT, and ultrasound, has greatly improved how we see into the body, and machine learning will allow us to identify earlier, and possibly more accurately, the findings of these modalities. However, seeing disease earlier might not lead to improved outcomes. (We should be particularly concerned about the early diagnosis of diseases not well understood and for which we have no meaningful treatments. If AI causes more lead-time bias, it might end up being an anxiety-creation tool more than anything else.) If we are to harness the power of AI, we should continually focus on how AI can improve clinical decision-making: our testing, differential diagnosis, and therapeutic decisions.

We have wonderful advanced technologies in US healthcare, but our outcomes are not even close to what they should be given how much we spend. AI, if used correctly, could provide us with the data and understanding to reduce waste inherent in inappropriate testing and therapy. On the other hand, we might fall in love with our technologic prowess of “seeing more,” and thereby “medicalize” our collective lives further, burdening our patients with information that leads to more “incidentalomas” and other iatrogenic disease. Ideally, a new model and new tools will evolve to improve decisions by both patients and doctors by using more accurate and personalized information.

So, in the interim, before we colonize Mars and before AI puts dinner on the table, let’s help the discussion by reducing the hype. Perhaps we should relabel the abbreviation of AI to mean “augmented intelligence,” not “artificial intelligence.” These words might really matter. Artificial intelligence suggests cyborgs and robots, making for good headlines but distracting everyone from the critical issues. Augmented intelligence speaks to guidance and aiding thinking. We should use AI to enhance skills and make sure not to “de-skill.” As an example in our work, we are augmenting the skin exam with machine learning by helping non-dermatologists to describe the skin exam features. A photo is analyzed by machine learning, but the physician must review and accept the analysis. This review likely has an educational effect, thereby increasing skill, as opposed to reducing it.

AI has the great potential to create more meaningful feedback loops in medicine by mining data over time and letting us see how individual outcomes match to population trends. Our patients will have access to rich data, new apps will evolve, and personal decisions will improve if we can build the critical feedback loops into a learning system. It is time we use big data and AI to understand what are the truly useful tests and therapies. The promise of AI is exciting, but let’s be careful not to forget the core problems that drive our national healthcare crisis. More technology is not synonymous with more common sense, and technology alone is not the answer to where we need to place our resources. As we use AI and machine learning to enhance care delivery for individuals, let’s also use AI to bolster public health decisions. We need to bring scientific rigor and meaningful data to the hard policy decisions that will be before us as the population ages and healthcare costs spiral further out of control.

Art Papier, MD, is the co-founder and CEO of VisualDx. A dermatologist and medical informatics expert, Papier is also an associate professor of dermatology and medical informatics at the University of Rochester School of Medicine and Dentistry. He is a thought leader in clinical informatics and healthcare solutions that improve diagnostic accuracy.


Tagged with:

About author

Related Articles