Free EMR Newsletter Want to receive the latest news on EMR, Meaningful Use, ARRA and Healthcare IT sent straight to your email? Join thousands of healthcare pros who subscribe to EMR and EHR for FREE!

Selecting the Right AI Partner in Healthcare Requires a Human Network

Posted on March 1, 2017 I Written By

Janae builds inbound social media sales and marketing plans for healthcare IT companies. Healthcare as a human right. Physician Suicide Loss Survivor. twitter: @coherencemed

Artificial Intelligence, or AI for short, does not always equate to high intelligence and this can have a high cost for healthcare systems. Navigating the intersection of AI and healthcare requires more than clinical operations expertise; it requires advanced knowledge in business motivation, partnerships, legal considerations, and ethics.

Learning to Dance at HIMSS17

This year I had the pleasure of attending a meetup for people interested in and working with AI for healthcare at the Healthcare Information and Management Systems Society (HIMSS) annual meeting in Orlando, Florida. At the beginning of the meetup Wen Dombrowski, MD, asked everyone to stand up and participate in a partner led movement activity. Not your average trust fall, this was designed to teach about AI and machine leaning while pushing most of us out of our comfort zones and to spark participants to realize AI-related lessons. One partner led and the other partner followed their actions.

Dedicated computer scientists, business professionals, and proud data geeks tested their dancing skills. My partner quit when it was my turn to lead the movement. About half of the participants avoided eye contact and reluctantly shuffled their feet while they half nursed their coffee. But however awkward, half the participants felt the activity was a creative way to get us thinking about what it takes for machines to ‘learn’. Notably Daniel Rothman of MyMee had some great dance moves.

I found both the varying feedback and equally varying willingness to participate interesting. One of the participants said the activity was a “waste of time.” They must have come from the half of the room that didn’t follow mirroring instructions. I wonder if I could gather data about what code languages were the specialty of those most resistant. Were the Python coders bad at dancing? I hope not. My professional training is actually as a licensed foreign language teacher so I immediately corroborated the instructional design effectiveness of starting with a movement activity.

There is evidence that participating in physical activity preceding learning makes learners more receptive and allows them to retain the experience longer. “Physical activity breaks throughout the day can improve both student behavior and learning (Trost 2007)” (Reilly, Buskist, and Gross, 2012). I assumed that knowledge of movement and learning capacity was common knowledge. Many of the instructional design comments Dr. Dombrowski received while helpful, revealed participants’ lack of knowledge about teaching and cognitive learning theory.

I could have used some help at the onset in choosing a dance partner that would have matched and anticipated my every move. The same goes for healthcare organizations and their AI solutions.  While they may be a highly respected institution employing some of the most brilliant medical minds, they need to also become or find a skilled matchmaker to bring the right AI partner (our mix of partners) to the dance floor.

AI’s Slow Rise from Publicity to Potential

Artificial Intelligence has experienced a difficult and flashy transition into the medical field. For example, AI computing has been used to establish consensus with imaging for radiologists. While these tools have helped reduce false positives for breast cancer patients, errors remain and not every company entering AI has equal computing abilities. The battle cry that suggested physicians be replaced with robots seems to have slowed robots. While AI is gaining steam, the potential is still catching up with the publicity.

Even if an AI company has stellar computing ability, buyers should question if they also have the same design for outcome. Are they dedicated to protecting your patients and providing better outcomes, or simply making as much profit as possible? Human FTE budgets have been replaced by computing AI costs, and in some instances at the expense of patient and data security.  When I was asking CIOs and smaller companies about their experiences, many were reluctant to criticize a company they had a non-disclosure agreement with.

Learning From the IBM Watson and MD Anderson Breakup

During HIMSS week, the announcement that the MD Anderson and IBM Watson dance party was put on hold was called a setback for AI in medicine by Forbes columnist Matthew Herper. In addition, a scathing report detailing the procurement process written by the University of Texas System Administration Audit System reads more like a contest for the highest consulting fees. This suggests to me that perhaps one of the biggest threats to patient data security when it comes to AI is a corporation’s need to profit from the data.

Moving on, reports of the MD Anderson breakup also mention mismanagement including failing to integrate data from the hospital’s Epic migration. Epic is interoperable with Watson but in this case integration of new data was included in Price Waterhouse Cooper’s scope of work. If poor implementation stopped the project, should a technology partner be punished? Here is an excerpt from the IBM statement on the failed partnership:

 “The recent report regarding this relationship, published by the University of Texas System Administration (“Special Review of Procurement Procedures Related to the M.D. Anderson Cancer Center Oncology Expert Advisor Project”), assessed procurement practices. The report did not assess the value or functionality of the OEA system. As stated in the report’s executive summary, “results stated herein are based on documented procurement activities and recollections by staff, and should not be interpreted as an opinion on the scientific basis or functional capabilities of the system in its current state.”

With non-disclosure agreements and ongoing lawsuits in place, it’s unclear whether this recent example will and should impact future decisions about AI healthcare partners. With multiple companies and interests represented no one wants to be the fall guy when a project fails or has ethical breaches of trust. The consulting firm of Price Waterhouse Coopers owned many of the portions of the project that failed as well as many of the questionable procurement portions.

I spoke with Christine Douglas part of IBM Watson’s communications team and her comments about the early adoption of AI were interesting. She said “you have to train the system. There’s a very big difference between the Watson that’s available commercially today and what was available with MD Anderson in 2012.”  Of course that goes for any machine learning solution large or small as the longer the models have to ‘learn’ the better or more accurate the outcome should be.

Large project success and potential project failure have shown that not all AI is created equally, and not every business aspect of a partnership is dedicated to publicly shared goals. I’ve seen similar proposals from big data computing companies inviting research centers to pay for use of AI computing that also allowed the computing partner to lease the patient data used to other parties for things like clinical trials. How’s that for patient privacy! For the same cost, that research center could put an entire team of developers through graduate school at Stanford or MIT. By the way, I’m completely available for that team! I would love to study coding more than I do now.

Finding a Trusted Partner

So what can healthcare organizations and AI partners learn from this experience? They should ask themselves what their data is being used for. Look at the complaint in the MD Anderson report stating that procurement was questionable. While competitive bidding or outside consulting can help, in this case it appears that it crippled the project. The layers of business fees and how they were paid kept the project from moving forward.

Profiting from patient data is the part of AI no one seems willing to discuss. Maybe an AI system is being used to determine how high fees need to be to obtain board approval for hospital networks.

Healthcare organizations need to ask the tough questions before selecting any AI solution. Building a human network of trusted experts with no financial stake and speaking to competitors about AI proposals as well as personal learning is important for CMIOs, CIOs and healthcare security professionals. Competitive analysis of industry partners and coding classes has become a necessary part of healthcare professionals. Trust is imperative and will have a direct impact on patient outcomes and healthcare organization costs. Meetups like the networking event at HIMSS allow professionals to expand their community and add more data points, gathered through real human interaction, to their evaluation of and AI solutions for healthcare. Nardo Manaloto discussed the meetup and how the group could move forward on Linkedin you can join the conversation.

Not everyone in artificial intelligence and healthcare is able to evaluate the relative intelligence and effectiveness of machine learning. If your organization is struggling, find someone who can help, but be cognizant of the value of the consulting fees they’ll charge along the way.

Back to the dancing. Artificial does not equal high intelligence. Not everyone involved in our movement activity realized it was actually increasing our cognitive ability. Even those who quit, like my partner did, may have learned to dance just a little bit better.

 

Resources

California Department of Education. 2002. Physical fitness testing and SAT9 Retrieved May 20, 2003, from www.cde.ca.gov/statetests/pe/pe.html

Carter, A. 1998. Mapping the mind, Berkeley: University of California Press.

Czerner, T. B. 2001. What makes you tick: The brain in plain English, New York: John Wiley.

Dennison, P. E. and Dennison, G. E. 1998. Brain gym, Ventura, CA: Edu-Kinesthetics.

Dienstbier, R. 1989. Periodic adrenalin arousal boosts health, coping. New Sense Bulletin, : 14.9A

Dwyer, T., Sallis, J. F., Blizzard, L., Lazarus, R. and Dean, K. 2001. Relation of academic performance to physical activity and fitness in children. Pediatric Exercise Science, 13: 225–237. [CrossRef], [Web of Science ®]

Gavin, J. 1992. The exercise habit, Champaign, IL: Human Kinetics.

Hannaford, C. 1995. Smart moves: Why learning is not all in your head, Arlington, VA: Great Ocean.

Howard, P. J. 2000. The owner’s manual for the brain, Austin, TX: Bard.

Jarvik, E. 1998. Young and sleepless. Deseret News, July 27: C1

Jensen, E. 1998. Teaching with the brain in mind, Alexandria, VA: Association for Supervision and Curriculum Development.

Jensen, E. 2000a. Brain-based learning, San Diego: The Brain Store.

Reilly, E., Buskist, C., & Gross, M. K. (2012). Movement in the Classroom: Boosting Brain Power, Fighting Obesity. Kappa Delta Pi Record, 48(2), 62-66. doi:10.1080/00228958.2012.680365.

The Value Of Pairing Machine Learning With EMRs

Posted on January 5, 2017 I Written By

Anne Zieger is veteran healthcare consultant and analyst with 20 years of industry experience. Zieger formerly served as editor-in-chief of FierceHealthcare.com and her commentaries have appeared in dozens of international business publications, including Forbes, Business Week and Information Week. She has also contributed content to hundreds of healthcare and health IT organizations, including several Fortune 500 companies. Contact her at @ziegerhealth on Twitter or visit her site at Zieger Healthcare.

According to Leonard D’Avolio, the healthcare industry has tools at its disposal, known variously as AI, big data, machine learning, data mining and cognitive computing, which can turn the EMR into a platform which supports next-gen value-based care.

Until we drop the fuzzy rhetoric around these tools – which have offered superior predictive performance for two decades, he notes – it’s unlikely we’ll generate full value from using them. But if we take a hard, cold look at the strengths and weaknesses of such approaches, we’ll get further, says D’Avolio, who wrote on this topic recently for The Health Care Blog.

D’Avolio, a PhD who serves as assistant professor at Harvard Medical School, is also CEO and co-founder of AI vendor Cyft, and clearly has a dog in this fight. Still, my instinct is that his points on the pros and cons of machine learning/AI/whatever are reasonable and add to the discussion of EMRs’ future.

According to D’Avolio, some of the benefits of machine learning technologies include:

  • The ability to consider many more data points than traditional risk scoring or rules-based models
  • The fact that machine learning-related approaches don’t require that data be properly formatted or standardized (a big deal given how varied such data inflows are these days)
  • The fact that if you combine machine learning with natural language processing, you can mine free text created by clinicians or case managers to predict which patients may need attention

On the flip side, he notes, this family of technologies comes with a major limitation as well. To date, he points out, such platforms have only been accessible to experts, as interfaces are typically designed for use by specially trained data scientists. As a result, the results of machine learning processes have traditionally been delivered as recommendations, rather than datasets or modules which can be shared around an organization.

While D’Avolio doesn’t say this himself, my guess is that the new world he heralds – in which machine learning, natural language processing and other cutting-edge technologies are common – won’t be arriving for quite some time.

Of course, for healthcare organizations with enough resources, the future is now, and cases like the predictive analytics efforts going on within Paris public hospitals and Geisinger Health System make the point nicely. Clearly, there’s much to be gained in performing advanced, liquidly-flowing analyses of EMR data and related resources. (Geisinger has already seen multiple benefits from its investments, though its data analytics rollout is relatively new.)

On the other hand, independent medical practices, smaller and rural hospitals and ancillary providers may not see much direct impact from these projects for quite a while. So while D’Avolio’s enthusiasm for marrying EMRs and machine learning makes sense, the game is just getting started.

Artificial Intelligence Can Improve Healthcare

Posted on July 20, 2016 I Written By

Anne Zieger is veteran healthcare consultant and analyst with 20 years of industry experience. Zieger formerly served as editor-in-chief of FierceHealthcare.com and her commentaries have appeared in dozens of international business publications, including Forbes, Business Week and Information Week. She has also contributed content to hundreds of healthcare and health IT organizations, including several Fortune 500 companies. Contact her at @ziegerhealth on Twitter or visit her site at Zieger Healthcare.

In recent times, there has been a lot of discussion of artificial intelligence in public forums, some generated by thought leaders like Bill Gates and Stephen Hawking. Late last year Hawking actually argued that artificial intelligence “could spell the end of the human race.”

But most scientists and researchers don’t seem to be as worried as Gates and Hawking. They contend that while machines and software may do an increasingly better job of imitating human intelligence, there’s no foreseeable way in which they could become a self-conscious threat to humanity.

In fact, it seems far more likely that AI will work to serve human needs, including healthcare improvement. Here’s five examples of how AI could help bring us smarter medicine (courtesy of Fast Company):

  1. Diagnosing disease:

Want to improve diagnostic accuracy? Companies like Enlitic may help. Enlitic is studying massive numbers of medical images to help radiologists pick up small details like tiny fractures and tumors.

  1. Medication management

Here’s a twist on traditional med management strategies. The AiCure app is leveraging a smartphone webcam, in tandem with AI technology, to learn whether patients are adhering to their prescription regimen.

  1. Virtual clinicians

Though it may sound daring, a few healthcare leaders are considering giving no-humans-involved health advice a try. Some are turning to startup Sense.ly, which offers a virtual nurse, Molly. The Sense.ly interface uses machine learning to help care for chronically-ill patients between doctor’s visits.

  1. Drug creation:

AI may soon speed up the development of pharmaceutical drugs. Vendors in this field include Atomwise, whose technology leverages supercomputers to dig up therapies for database of molecular structures, and Berg Health, which studies data on why some people survive diseases.

  1. Precision medicine:

Working as part of a broader effort seeking targeted diagnoses and treatments for individuals, startup Deep Genomics is wrangling huge data sets of genetic information in an effort to find mutations and linkages to disease.

In addition to all of these clinically-oriented efforts, which seem quite promising in and of themselves, it seems clear that there are endless ways in which computing firepower, big data and AI could come together to help healthcare business operations.

Just to name the first applications that popped into my head, consider the impact AI could have on patient scheduling, particularly in high-volume hostile environments. What about using such technology to do a better job of predicting what approaches work best for collecting patient balances, and even to execute those efforts is sophisticated way?

And of course, there are countless other ways in which AI could help providers leverage clinical data in real time. Sure, EMR vendors are already rolling out technology attempting to help hospitals target emergent conditions (such as sepsis), but what if AI logic could go beyond condition-specific modules to proactively predicting a much broader range of problems?

The truth is, I don’t claim to have a specific expertise in AI, so my guesses on what applications makes sense are no better than any other observer’s. On the other hand, though, if anyone reading this has cool stories to tell about what they’re doing with AI technology I’d love to hear them.