Monday, November 2, 2015

Summary of Intro to SciKit Learn at Open Data Science Conference

Lukas Biewald, CEO of Crowdflower, gave a three hour workshop on arguably the most vital module to machine learning – SciKit Learn. Lukas gave the audience in Microsoft’s Hack Reactor Space open source code to follow along from:

He highlighted the value of participating in Kaggle competitions. He defined regression as one variable trying to predict another variable. He gave the students in the audience the main project for today’s workshop – judging emotion about brands and products. In this project, we saw Tweets about several Apple brands and products.

99% of the time was spent on “should we drop I” or “should we drop my” types of questions. We also asked “should we remove really rare words?” “should we remove really common words?” “should we remove stop words?” and “what is a word?”

In the midst of our learning, Mr. Biewald purported that if one did not want to earn a degree in machine learning, memorize the map at this website:

When testing how well an algorithm was working, Lukas recommended we utilize the “” program. Moreover, that we should not be testing on the data we are training on. For example, we should generally train on 80% of the data and test on 20% of the data.

On the “” program, he used the chi2 statistical model to rank which column was the best column in our datasets. However, he claimed that the chi2 feature was the least important to consider of all of the topics we discussed today.

Lukas closed with a call for us to make our own features like a count for the number of exclamation points, emojis, lengths of Tweets, or language of Tweets.

Predictive Analytics Summary from Open Data Science Conference

Don Dini (AT&T, USA) opened his talk on Predictive Analytics with a comment on how machine learning was the second best solution to every problem. He made a self-contained pun on the word TERRIFYING being equivalent to:

[‘T’, ‘EH1’, ‘R’, ‘AH0’, ‘F’, ‘AY2’, ‘IH0’, ‘NG’]

He continued with a generalization that is the general answer was complicated, data science was generally the answer.

If you believe your servers are being attacked, perform hypothesis testing using null distribution KDE fromed from the last month of data. Then propose a model capable of addressing the inference problem. Finally, utilize k-nearest neighbors linear regression support vector machines.

As an example, Mr. Dini gave us a simple “Predict the next number” problem. To solve this, suppose
xi ~N(mu, sigma2)
a.     Then µ - hat = (x1 + x2 + … + xn) / N
b.     Then the coder needs to evaluate how well the model did. How sure is the model about what is recommended.
c.     Then determine computational confidence intervals

To perform computational confidence intervals, “bootstrapping” was a helpful method. Consider having 3 similar data sets:
a.     x1, x2, …, xn à r(x)
b.     x1, x2, …, xn à r(x)
c.     x1, x2, …, xn à r(x)
Ideally, if a coder had a perfect model, and the data the coder had was perfect, the outcome should be the same for all models. Moreover, the amount of data present was a direct (and non-removable) cause of uncertainty in the prediction. In any prediction made, it was vitally important to communicate uncertainty. For example, Don tried to show us how he answered the question “What are the things that influence communication on social media?”
For example, if thirty seconds had elapsed from a user’s Tweet, how much longer until that same user would Tweet again? Don’s point was that uncertainty, here, made a real prediction less meaningful.

Continuing on to the next point, Don highlighted the next relevant problem “How do we know if two variables had anything to with each other?”

Consider: Y = x2
x: x1, x2, ..., xn
y: y1, y2, …, yn

In the above instance, covariance decreased as odd-degree polynomials increased in degree. Conversely, the same result occurred for all even degree polynomials.

The next topic Don shared with the Microsoft Hack Reactor Office was Entropy (a principle that explained how decision trees worked).
X took on a series of values
                  x: x1, x2, …, xk
                  Pr(x1), Pr(x2), …, Pr(xk)
                  H(x): Pr(x1) * log(Pr(x1))
x also took on a minimum value of 0 when it was most skewed:
[0, 0, …, 0, 1, 0, …0, 0]
x took on a maximum when it was least skewed:
[1/k, 1/k, … 1/k]
Don explained that we could sort variable by how much they could cause x’s entropy to decrease. The goal was to find the variable that was going to make him maximally certain. This feat required defining relative entropy as:
H(X|Y) = Pr(y = y1) * H(X|Y = y1) + Pr(y = y2)

If Don had a collection of variables: A, B, C, …, he could sort them out by how much they caused entropy to drop:
a.     H(X) – H(X|A)
b.     H(X) – H(X|B)
c.     H(X) – H(X|C)

This brought Don to his next principle, “Mention Distance.” Rather, the amount of seconds that had elapsed since someone had Tweeted. With mention distance, Don was trying to answer the following question: “If I knew how long it had been since someone in this front network had mentioned them, did it influence if that person would respond?”

Don closed with a group activity to have us practice the boostrapping method.

Wednesday, October 28, 2015

R Summary from Open Data Science Conference

Gabriela de Queiroz (Sharethrough, USA) kicked off the first workshop of the weekend with an Introduction to R. She shared that the following 3 installation packages within R were very useful: ggplot2, tidyr, and dplyr. She continued to state that everything is a data frame and that if you’re familiar with SQL, dplyr is very similar.

Ms. De Queiroz also went over how to summarize data using the following 3 methods:
i.                summarise(data, avg = mean(col_a): summarises data into a single row of values.
ii.              Summarise_each(data, funs(mean)): applies summary function to each column
iii.             count(data, vars_to_group_by, wt = col_a): counts the number of rows with each unique value of variables (with or without weights).

One of the more important concepts she introduced was the PIPES concept. Gabriela purported that pipes make one’s code more efficient and legible. It utilizes the characters: [filename]%>%.

Miss. De Queiroz continued with a tutorial on using the ggplot2 package installation from R to visualize data. To build a ggplot, the coder needs to:
i.                bind the plot to a specific data frame using the data argument
a.     ggplot(data = [filename]
ii.              define aesthetics (aes) that map variables in the data to axes on the plot or to plotting size, shape, and color.
iii.             Add gemos
a.     ggplot(data = [filename], aes(x = dep_delay, y = arr_delay))
b.     gemo_point(alpha = 0.1, na.rm = TRUE)
c.      Add colors

d.     Add a boxplot

Tuesday, October 6, 2015

Stanford Medicine X | Ed Day 1

Introduction to Stanford Medicine X | Ed

A MedX hangover kit, educational learning labs, 3D printing, and the famous Zoe Chu, the dog – these were all aspects of Stanford Medicine X | Ed 2015, as envisioned by the creative mind of conference founder and executive director Larry Chu, M.D. As I listened to the myriad of student and teacher experiences ranging from “learning module recommendation engines” to “social media’s roles in patient safety”, I realized how unique Stanford Medicine X | Ed was. I also realized how there was no medical education innovation conference like it in the world.
Dr. Chu opened with comments on how patients and their caregivers were the biggest untapped resources in healthcare. He wanted everyone in the audience to focus on five major themes: 1) Acknowledging the unique needs and challenges millennial learners are experiencing to pursue their education. 2) Focus on how to scale 3) See how social media and tools serve as a new classroom 4) Explore interdisciplinary cooperation to improve clinical care 5) Lifelong learning.

Stanford Medicine X |Ed

Howard Rheingold (Stanford University, USA) kicked off the first day of Stanford Medicine X | Ed with a talk on the value of the networked patient (i.e. e-Patients). He discussed five pitfalls of networked patients: 1) Technology and the way people use it had a lot of different sides to it 2) Patients relied on false medical information found on the internet 3) Overconfidence in the internet 4) Quality of “crap detection” 5) Guiding e-Patients takes time. Conversely, Mr. Rheingold also highlighted the opportunities of the e-Patient industry: 1) Learning was about knowledge, not technology. 2) No individual would keep up with every development in a particular disease as well as a well-informed network could. 3) Learning was about relationships, not technology.
Next up was Joseph Santini (Gallaudet University, CA) giving a talk on Autonomy, Access, and Patient Engagement via American Sign Language. He talked about how deaf medical students were denied access to basic healthcare needs. Moreover, remote access to healthcare providers did not equate to autonomy. Joseph closed with the three requirements of autonomy: 1) Engagement with the patient 2) Education 3) Ensurance. Ryan Hanes (Osmosis, USA) spoke on the unmet needs of medical learners and how to address them. He focused on the problem with human memory associations – they associated one word with only one memory. In reality, memories were associated with multiple associations. To reinforce medical principles, Osmosis takes real celebrity patient cases. For example, Osmosis explained how Serena Williams’ dry eyes from her Sjogren’s Syndrome affected her tennis game.
Dhruv Khullar, M.D. (Massachusetts General Hospital, MA) spoke briefly on the importance of sitting with patients. Dr. Khullar purported that the most draining aspect of medical education was not long hours, rude colleagues, or steep learning curves. It was taking the time to sit with the patient and see things from their perspectives.
Nisha Pradhan (De Anza College, Pennsylvania) spoke on the unmet needs of today’s pre-medical learners and the need to engage early and often to nurture next generation providers. Ms. Pradhan spoke about the importance of teaching pre-med students critical thinking skills instead of memorization techniques. She mentioned how she has on the MCAT more psychology and sociology related questions requiring open-minded critical inter-disciplinary thinking.
Shiv Gaglani (Osmosis, USA) commented on how medical schools could be improved by transitioning to a competency based model from a time-based model. Overall, the panel agreed that the new millennial learners are creative and innovative.

Medical Educators – Opportunities and Challenges for Innovation

The day continued with another student speaker, Tim Van de Graft, M.D. (University of Amsterdam, NLD) gave an Ignite! talk on how to re-shape medical education with empathy in the Netherlands.
Dave deBronkart (ePatient Leader, USA) started his talk about the importance of being able to screen for quality medical content on the Internet. He brought to Plenary Hall’s attention that 1/20 Americans are misdiagnosed every year. To demonstrate the importance of the engaged patient, he cited a case where doctors apologized for telling a 19-year-old cancer patient to ‘stop Googling’ symptoms – she died. Charles Prober, M.D. continued the discussion with an emphasis on bringing patient-centered education to basic science courses. He closed with a list of six benefits of a new educational paradigm: 1) Consensus-driven content creation 2) Efficient & Widespread dissemination 3) Curricular flexibility 4) Locally adapted 5) Broadened faculty participation 6) Enhanced student engagement.
Michael Evans, M.D. (University of Toronto, CAN) introduced the motto “we suck less incorporated” to Stanford Medicine X. Dr. Evans re-iterated it as “we failed a lot.” He analogized his company to that of a urinal stall that has nudges for the urine to flow down the stall without splattering. Each product the company failed with counts as one nudge – more nudges mean a  better functioning stall.
Anne Marie Cunningham (Cardiff University, UK) moderated a panel of the aforementioned medical educators. Adeel Yang (Picmonics, USA) thanked his school (University of Arizona) for supporting his efforts to create Picmonics by “staying out of his way.” Professor deBronkart challenged that not every innovation was disruptive. Dr. Prober exemplified Gilbert Chu, M.D., Ph.D.’s molecular biology course as the perfect example of patient-centered education because he starts each lecture off with a patient’s story.

Social Media, Collaboration + Open Access: The New Classroom

Marie Ennis O’Connor (BCCEU, IRL) spoke about social media as a tool for patient-driven medical education. Her key point was that learning was enhanced by social cognitivity. Stuart Haines, Pharm.D. (University of Maryland, USA) spoke on social media in pharmacy education. He described the tool he made where Twitter was used to page pharmacy students and the students would call using Google Voice. Ajay Major, M.B.A. (Albany Medical College, USA) highlighted his experience helping with “The White Coats for Black Lives Die In” to garner 28,000 participants across the country. Bertalan Mesko (The Medical Futurist, HUN) spoke about his new book “My Health Upgraded.” He also gave his views on the five necessary components of providing a cutting edge medical curriculum: 1) Engaging presentations with the most exciting up to date information 2) e-Learning platform 3) Facebook challenge 4) Evaluations 5) Chocolate-based rewards.
Bryan Vartabedian, M.D. moderated the panel on social media and the new classroom. Rob Rogers, M.D. mentioned how this is the first time he had ever thought about involving patients into the medical conversation in his life. Ajay poked fun at how one of his colleagues teaches students by posting educational rap videos. Dr. Haines brought up a sobering point that 1st year student-pharmacists were not ready to contribute to a professional setting, only to a social space like Facebook Groups or Instagram posts.

Update from Singapore: Challenges and Opportunities for Innovation in Medical Education in an Asian Context

Naomi Low-Beer (Lee Kong Chian School of Medicine, SGP) gave Medicine X| Ed an update on the challenges and opportunities for innovation in medical education in an Asian context. She also recognized that there was more to medical education than identifying with medicine.

Update on Medical Education from Europe

Anne Marie Cunningham, M.D. gave plenary hall an update on innovation in medical education in Europe. She highlighted the First International Medical Education Hackathon where medical students, learners, and coders tried to get together to develop new projects. She also brought to our attention The Tuning Project and a problem-based learning MOOC.

Interdisciplinary Learning

Moving on to the 4th and diversity-friendly theme of the conference: interdisciplinary learning. Emily Kramer-Golinkoff regailed the crowd with a story on her battle against cystic fibrosis. Ms. Kramer-Golinkoff was an orphan of the disease, holding two non-sense copy mutations resulting in cystic fibrosis. Paul Haidet, M.D. (Penn State Hershey Internal Medicine, USA) opened his talk with a rendition of the Waltz Trio. He analogized the roles of the pianist, bassist, and drummer to those of a pharmacist, physician, and nurse. Heather Davidson, Ph.D. (Vanderbilt University, USA) highlighted her work with the Vanderbilt Program in Interprofessional Learning (VPIL). VPIL trained medical, nursing, and pharmacy students to work collaboratively. The program’s goals were to: 1) Cultivate 2) Prepare 3) Improve 4) Integrate 5) Nurture. Tina Brock, Ph.D. (University of California San Francisco, USA) shared here experiences providing clinical services in Sub-Saharan Africa. There, the region’s top two priorities were healthcare and education. Comparing Namibia’s meager pharmacist staff size of 350 to California’s 25,000 (Namibia’s country is twice the size the state of California), she brought to our attention Namibia’s pharmacist staff shortage.
Kevin Clauson, Pharm.D. moderated the panel of leaders on how to overcome the challenges in the field of interdisciplinary learning. Dr. Haidet chimed that unless nurses, physicians, and pharmacists could see things from each other’s point of view, they could not “play music together.” Ms. Kramer-Golinkoff proposed to crowdsource the issues we were experiencing. Moreover, she highlighted the importance of learning about what her friends were doing to solve the same cystic fibrosis issues she was experiencing. Dr. Low-Beer purported the importance of identifying a curriculum that was authentic.

Longitudinal and Lifelong Learning: The Future of CME

Leslie Rott, Ph.D. (Sarah Lawrence College, USA) gave a talk on her journey of lifelong learning as a patient with chronic illness. Paul Teirstein, M.D. Paul S. Teirstein, M.D. (Scripps Health, USA) shared his work that concluded physicians learned best through professional Continuing Medical Education courses. Brian McGowan, Ph.D. (ArcheMedX, Inc., USA) presented the results of his Continuous Learning research project with three aims: 1) Understand what actions millennial learners wanted to take 2) Understand how and when the actions were used 3) Verify if those actions were effective. Dr. McGowan drew four conclusions on effective actions to improve learning: 1) Note-taking 2) Setting reminders 3) Searching 4) Externally nudging on important points.
Kirsten Ostherr, Ph.D. (Rice University, USA) moderated the panel on the future of continuing medical education composed of Dr. Haidet, Dr. McGowan, Jamia Crockett, Dr. Cunningham, and Dr. Teirstein. Dr. Teirstein highlighted the importance of experiential design. Brian McGowan expounded upon a controversial opinion – just because a student scored great on the MCAT did not mean he was a good learner. Ms. Crockett spoke about the role patients needed to play in medical education. With regard to collection patient feedback on doctor visits, Anne-Marie said we needed to look at how many people were “Yelp’ing” their physicians.

Closing MedEd Keynote Day I:

Abraham Verghese, M.D. (Stanford University, USA) closed the opening day of Medicine X|Ed with a spectacular keynote. His first point was on how our teachers made us who we were as professionals. He moved on claim that doctor burnout was due to physicians spent most of their time on the computer. Unless physicians develop a ritual of actually being with the patient, no amount of time on the computer would prevent “short-changing” our patients. He also claimed no YouTube tutorial could teach physicians to develop an “engaged patient” ritual. To teach these skills required daily mentorship over years of time.

Wednesday, June 3, 2015

WLSA Convergence Summit 2015


A report on The 10th Annual Wireless Life Sciences Alliance Convergence Summit, held at the Omni Hotel, San Diego, CA, U.S.A., May 26th-28th, 2015.

Meeting Report

Graphic listeners, virtual iPhone screens on your forearm, and Apple Watches, these were just a few of the highlights of the 10th Annual Convergence Summit. As Robert McCray, J.D. (Wireless Sciences Alliances, U.S.A.) opened with comments on how we were in a phase of digital health where only the most committed companies were involved to create better solutions to bigger healthcare problems. He also highlighted the importance of having no emotional connections to healthcare, analogizing his point with John Waters’ work in rock and roll music. Eric Topol, M.D. (Scripps Translational Science Institute) highlighted the advent of an iPhone hologram on your forearm that had all of the functionalities of an actual iPhone. The only piece of hardware that was required to use it was a miniature wristband. He closed his keynote with a comment on how doctors were losing a fighting battle for hospital control to nurse practitioners. The 10th Convergence Summit integrated issues in regulation, partnerships with Scandinavians, and connected health care’s role in hospitals to update the community on the current state of the Digital Health Industry.

Digital Diagnostic and Therapeutic Platforms
Lisa Suennen (Venture Valkyrie Consulting, U.S.A.) moderated the first panel of the Summit. She brought attention to the problem of clinicians diagnosing patients with diseases for which there were no treatments – Parkinson’s Disease and Alzheimer’s Disease. Ron Gutman (HealthTap, U.S.A.) took to the stage with comments on crowdsourced medicine and how wearables were re-defining the “SOAP’ing” process clinicians used to evaluate the medical profiles of their patients. Moreover, HealthTap is standardizing medicine by providing access to the clinical advice of tens of thousands of doctors.
Brian Otis, Ph.D. (Google, U.S.A.) urged the digital health community to have an unrelenting respect for the regulatory process. Continuing on, Euan Thomson (Khosla Ventures, U.S.A.), iterated that if we had Uber in medicine, medicine would be simplified.
Dr. Topol exemplified how Switzerland had a health bank for each of their citizens and that other countries needed to follow suit.
He closed with a comment on the ability to monetize patient data was a valuable skill. The problem was making sure you did not violate patients’ rights to access said data.

Making Consumer Driven Healthcare a Reality – The Opportunity and Regulatory Realities
Up next was Ralph Simon leading a discussion on how to overcome regulatory issues holding consumer driven healthcare from becoming a reality. In the panel included Douglas McClure (FitLinxx, U.S.A.), Rob McCray, Kim Tyrrell-Knott (Epstein Becker & Green, P.C., U.S.A.), and Harry Wang (Parks Associates, U.S.A.). In terms of regulatory affairs, Ms. Tyrrell-Knott discussed the difficulties of defining the fine line between a regulated medical device and an unregulated hi-tech device. Rob called for a move toward implementation of digital health technologies in automobiles. Moreover, he made a notion that patient engagement techniques needed to be incorporated into the design of medical products. Harry highlighted his firm had seen a 24% increase in device adoption over the previous six quarters. The panel closed with a comment on how more effective digital solutions needed to be done in the mental health space. Overall, the panel agreed that in a consumer-driven healthcare system, the healthcare community needed to shift from a fee-for-service based system to a value-based system.

Connected Health Care for Hospitals
Don Jones moderated the following panel consisting of Marty Miller (Advanced ICU Care, U.S.A.) and Yulun Wang (InTouch Health, U.S.A.). Mr. Jones opened with a story on how nurses were the only healthcare professionals allowed to operate a robotic neurosurgeon at Walter Reed Medical Center. Moreover, a lot of digital health work in the hospital was being devoted to improving coordinated care teams. Yulun gave some eye-opening statistics he learned as the American Telemedicine Association President. 500,000 patients were taken care of by a tele-ICU physician. 125,000 were taken care of by a tele-stroke physician. 800,000 online consultations were done in the previous year. Mr. Wang wanted to make it clear that a tele-health physician consult was not necessarily a complete physician consult. Continuing on, Mr. Miller put the $2.5 billion market of serving the ICU into perspective by stating that statistic. He closed with a comment on how advanced practice practitioners covered multiple hospitals.
The Investor Panel
Julie Papenek (Canaan Partners, U.S.A.) led the next panel consisting of Casper de Clerq (Norwest Venture Partners, U.S.A.), M. Wainwright Fishburn (Cooley, LLP, U.S.A.), Lucian Iancovici, M.D. (Qualcomm Ventures, U.S.A.), and Maxwell Kahn (Merck GHI Fund, U.S.A.). Wainwright Fishburn opened with a comment on how digital health was a battle of the game: “Survival of the fittest.” Also, venture capital companies in the past avoided consumer-centric companies due to how fickle the consumer was. 
The concept of having patients more engaged in their own health is not financially helpful to all consumer-centric digital health companies. Dr. Iancovici added his input around the notion that more companies would build technology around drugs in the future. On investing, he mentioned the importance of focusing on companies involved in population health management. Mr. Kahn’s take on population health management was also focused on drugs. He purported that managing hepatitis C virus patients on triple therapies was more difficult than Diabetic patients on monotherapy. 
Moreover, Dr. Iancovici made a point that it was actually good to invest in entrepreneurs that take investors’ advice only 20% of the time. Lucian closed his contributions with a generalization to what was holding back digital health companies from succeeding: traditional hi tech companies did not understand the traditional healthcare business models. On cost-extraction, the panel believed population health management was universally beneficial to all of the stakeholders.
The panel closed with a joke on how they had all underestimated how successful FitBit was going to be.

Daniel Kraft Keynote
Daniel Kraft, M.D. (Singularity University, U.S.A.) ended Day 1 with an excellent keynote covering a wide array of topics summarizing the state of the healthcare innovation space. Dr. Kraft highlighted the use of a diaper for infants that sent messages whenever the child has soiled the diaper. He also focused part of his talk on the use of brain-computer interfaces. Moreover, he opined that gaming could be powerful in improving cognitive disabilities with multi-tasking. He also predicted the advent of robotic anesthesiologists and the prevalence of augmented intelligence in healthcare. Continuing on, he reported on people diagnosing their own brain tumors with the help of 3D printing.

The Digital Life Sciences
The second day of the Convergence Summit started off with Don Jones leading a panel composed of Rena Rosenberg, Ph.D. (Pharma & Medical Products Practice, U.S.A.), Jordan VanLare, M.D., Healthcare Systems and Services Practice, U.S.A.), Mike Doherty (Hoffmann-La Roche Ltd., C.H.E.), Romain Marmot (Voluntis, F.R.A.), and Andrew Obenshain (Sanofi, F.R.A.). Mr. Jones also wanted to highlight the importance of the Scripps Translational Science Institute’s Digital Medicine Group, led by Steven Steinhubl, M.D., and how he was heavily investing in its growth. Drs. Rosenberg and VanLare wanted to highlight 5 fundamental trends in the digital life sciences today: 1) Patient Behavior was changing 2) Government agencies were moving quickly 3) Trial data was necessary 4) Competition was faster and fiercer 5) Care was changing. On improving the negative image that pharmaceutical companies had, Dr. VanLare claimed that image would have improved if Big Pharma involved engaged patients in the design process of their novel technologies. Dr. Rosenberg closed with a notion that device design (like those from Medtronic) tended to be disconnected from the payor system, leaving patients and providers to suffer the expense. Mr. Doherty made comments about how data analysis was unaddressed by regulators today. For example, there was little regulation in the $60 million deal made between 23AndMe and Genentech’s 3000 patient’s genetic profiles. Romain highlighted an important problem amongst his one million insulin users: 50% did not reach their blood glucose goal and 25% stopped using their insulin. Moreover, if pharmaceutical companies were going to invest in solutions that supported the discussion of health outcomes based payments, accountable care organizations would only be a temporary solution. Mr. Obenshain highlighted how Sanofi was making more efforts to provide integrated care.

The International Discussion
Up next was the highly anticipated panel garnering the opinions of world-renowned international experts. Ralph Simon (Mobilium Global Ltd., U.K.) led the discussion with Peter Cowhey, Ph.D. (U.C.S.D., U.S.A.), Jack Kreindler, M.D. (Sentrian, U.S.A.), Scott Lambert (Ascension Health, U.S.A.), Anne Lidgard, Ph.D. (Vinnoa, S.W.E.), and Jordi Serrano Pons, Ph.D. (World Health Organization, C.H.E.). Mr. Cowhey highlighted how Beijing and Shanghai were fighting to be dominant in the technology sector. However, Shanghai was losing because Beijing was pouring large sums of research & development money indiscriminately. Shanghai tried to take the lead by looking to San Diego for help in the research & development department. On the use of machine learning and biosensors in data analysis, Dr. Kreingler claimed there was a data triangle between the United States of America, the United Kingdom, and the National Health Service in the U.K. Dr. Pons brought to attention the multitude of projects aiding infants in Africa. More specifically, a non-government organization called Zero Mothers Die that provides women with pay phones. Dr. Lidgard talked about her Stockholm based company relocating her in Silicon Valley to take advantage of the ecosystem’s resources. To add to her story, she agreed that Silicon Valley would not be the only innovation-based sector in the United States of America. For example, Chicago and New York were also hot spots.

The Hospital Discussion
The final panel of the Summit closed with Glenn Steele, M.D., Ph.D. (Geisinger Health System, U.S.A.), M. Michael Shabot, M.D. (Memorial Hermann Healthcare System, U.S.A.), P. Martin Paslick (HCA, U.S.A.), and Rosemary Kennedy, Ph.D., R.N., M.B.A. (Sotera Wireless, U.S.A.). Dr. Steele highlighted the factors technology companies needed to consider to change provider behavior: 1) Culture 2) Staffing 3) Data 4) Best Practices 5) Clinical Workflow 6) Clinical Decision Support. Dr. Kennedy brought to attention the fact that 36% of nurses spent time running up and down the hallway. Michael dishearteningly had to reveal that his health system had to block Apple iPhones and its upgrades because HCA’s network died using them. Moreover, he stated that HCA’s system was not mature enough to adopt speech-to-text or relationship extraction programs. Dr. Steele closed on a note stating that unless we were able to change the thinking of nurse practitioners, pharmacists, and physicians, the platform adoption and data analysis innovations hospitals were seeking to attain would never come to fruition. Moreoever, scaling an idea was not the same as innovating.

Breakout Sessions
For the first time in recent years, the Summit held breakout sessions for the attendees to gain expert opinions on contemporary issues facing the digital health industry. Steven Steinhubl, M.D. (Scripps Translational Science Institute, U.S.A.) focused on how to develop the evidence base to drive forwards the dissemination of clinical technologies. Jeff Belk (Velocity Growth, U.S.A.) held a session on crowdfunding. Kim Tyrrell-Knott gave a round table on regulatory issues surrounding clinical decision support. Eric Milch (Cooley, LLP, U.S.A.) gave a talk on innovations in patent law. Moshe Engelberg (Research Works, U.S.A.) talked about his work in providing local companies resources to connect with one another. 
Dr. Steinhubl mentioned that 2/3 of current healthcare professionals would not have recommended a healthcare profession to their children. Moreover, 1/3 of current healthcare professionals wanted to get out of the healthcare profession. In support of Dr. Glen Steele, Dr. Steinhubl agreed that the financial incentives to order unnecessary tests were perverse. For example, he had a patient that underwent a CT Scan, CABG, and a drug eluting stent the patient was not indicated for. Moreover, he had patients ask him to start medications like Xarelto because of an advertisement the patient saw. The moral dilemma? The drug was not the most appropriate choice for that patient. 

Unlike the theme at the 9th Annual WLSA Convergence Summit, the benefits of engaging more patients to utilize connected health devices were controversial among key opinion leaders this year. The rise of international adoption of the digital life sciences was one important takeaway from the 10th Annual Convergence Summit. With JSR from Japan, The Children’s Foundation of California outside of Tijuana, and Nordic Connected Health Start Track in Scandinavia, more international players demonstrated the global impact digital health was having in healthcare. Finally, there was not a clear consensus amongst key opinion leaders on whether patients should have sole ownership of their connected health or genetic data. Digital health was still in its nascent stages, however, some believed we were in a phase where only the most committed companies were still in this field.


Nicholas T. Vu, Pharm.D.

Twitter: @nicholasvu

World of Watson 2015 Summary

A report on the 1st Annual IBM World of Watson Conference, held at the Duggal Greenhouse, Brooklyn, New York, May 4th – 6th, 2015.

Meeting Report
Artificial Intelligence-derived recipes, Water Taxis, and a $25,000 hackathon, these were all aspects of IBM’s creative genius that The Duggal Greenhouse warmly welcomed. As I listened to the myriad of applications The IBM Watson had ranging from finance to medicine, I realized what made World of Watson unique from all other conferences. Cognitive computing science had officially hit consumers. This was the first consumer-oriented cognitive computing science conference that demonstrated practical value to the consumer across an array of different industries. Mike Rhodin (IBM, USA) said the Watson Team was addressing the fundamental problem of the ability to scale human knowledge. Ginni Rometty (CEO, IBM, USA), talked about how 250 independent software vendors were re-selling cognitive applications. Moreover, 5000 companies were in the pipeline for partnership. 100 universities were building the next round of entrepreneurs. To set the tone of the conference, Ms. Rometty highlighted three points for Watson’s cognitive services: 1) It could augment the decisions humans made 2) IBM really wanted an ecosystem where Watson was an open platform 3) There was no right and wrong. She closed her opening remarks with a powerful statement on how every decision man-kind made would eventually be informed by a cognitive system similar to Watson – and all human lives would be transformed because of it.

Watson Health
We would help oncology with 300 medical journals, 200 textbooks, 23 million articles, and every clinical trial ever done.
At World of Watson, they announced a number of new projects, including the Watson Genomic Analytics program and its role in matching DNA with tumors. Moreover, they announced the Alchemy API and the different APIs with services such as personality insights, speech to text function, machine translation, and others. Watson’s ability to process natural language and augment decisions made it novel from any other program prior to it. Lukas Wartman, M.D. (McDonnell Genome Institute at Washington University in St. Louis, USA) talked about his battle with leukemia and how Dr. Watson was essential for coming up with a sequence-based treatment plan. Norman Sharpless, M.D. (University of North Carolina Lineberger Cancer Center, USA) claimed that drugs programmed into Watson would have the ability to pick the humans they were appropriate for in the next two to three years. Watson was also learning to speak new languages. Annette Bruls, VPGM (Medtronic, Switzerland) closed with a talk on how over 300 million patients were diagnosed with Diabetes and how that number was rising quickly.

Transforming Industries and New Eras Cognitive Computing Opens
Mr. Rhodin made a comment on how each of person would generate 1100 terabytes of data. 20-30% of that data would be genomic data whereas the remainder of it would be exogenous data. With IBM Watson Health’s partnership involving Apple, Johnson & Johnson, and Medtronic, the team would be moving forward quickly.
Bridget van Kralingen (DBS, Singapore) talked about their digital banking services across 70 markets in Asia in countries including Taiwan and Singapore. She also talked about how scalability was a serious issue. Ms. Kralingen closed with a statement on how she would have loved to have a relationship manager via Watson. Moreover, the relationship manager program should have had the ability to like, dislike, comment, and click.
Stephen Gold (IBM, USA) continued the conversation with a segment on how his bike shopping experience would be transformed. He started with a story about how he got 270 million search results when he Google’d potential bicycles to purchase. To solve this problem, Brian O’Keefe (Sellpoints, USA) took the stage.
Mr. O’Keefe started with a comment about how Stephen’s Google’ing habits represented 60% of all shoppers. He also mentioned how Stephen’s probability of actually buying a bicycle was 30% if Mr. Gold differentiated between choosing a road category and a traditional category. Brian continued on to talk about how Stephen also preferred to test ride a bike before buying it. This shopping behavior brought to light another problem many shoppers wanted to avoid -- interacting with clerical staff that had no knowledge about the product the consumer was looking for.
Mike Garel (CEO, EyeQ) addressed the clerical staff issue by discussing the advantages of e-commerce. EyeQ’s use of e-commerce increased the company’s sales by 20%. However, 92% of the brand’s revenue still came from purchases in the physical store. EyeQ even gamified the shopping experience and took into their consumer’s personalities to customize their shopping experiences. To customize the experience, EyeQ leveraged their consumers’ Twitterfeeds, age, and gender.
Continuing on, Dan Hartveld (Redant, United Kingdom) continued to solve two major problems: 1) 57% of sales assistants received less than two hours of training before being put in front of a customer 2) 43% lied to customers every week due to a lack of project knowledge. His solution involves utilization of one of many APIs Watson offered: the Q&A API.
Stephen Gold re-directed the conversation to how Ford was re-inventing the car-shopping experience in a similar fashion to the bicycle industry. To discuss details, Marc Fecker (VP, Forddirect) came to the stage. Mr. Fecker brought to light how his company was trying to solve the “pushy salesman” problem. However, creating this experience and using Watson to re-train Ford’s sales staff had proven its own set of challenges.
On a panel involving Jon Iwata (IBM, USA), Chef James Bricione (Institute of Culinary Education, USA), Mark Kris, MD (Memorial Sloan Kettering Cancer Center, USA), and Beverly Olivier (Deakin University, USA), they discussed the their personal experiences using cognitive applications. Mr. Iwata initiated the conversation with a question on whether Watson was an icon of innovative business. Chef Bricione commented on his use of Watson for three years to formulate recipes that excited the chefs that worked for him. Dr. Kris highlighted how he knew IBM Watson was helpful when it took such a short amount of time to recommend appropriate chemotherapy regimens for his lung cancer patients. Finally, Ms. Olivier highlighted Watson’s use in the academic engagement sector in Australia. There were 50,000 students across Australia. 20% of them never came to campus. Regardless, every student and staff member could have asked Watson for advice. To help with Watson user engagement, Mr. Olivier went as far as to employ some students just to use Watson and give feedback.
Mike Rhodin and Chris Anderson (TED, USA) closed the first day with an exciting announcement on how IBM Watson would partner with TED’s 25,000 users. For example, this partnership would leverage the power of Watson’s Relationship Extraction API to find relevant videos based on the 1000s of TED talk videos posted.

I: Feature Presentation I: Data Analytics in Cognitive Computing
Terry Jones (Wayblazer, USA) started the morning off with a blurb on how we cannot predict the future applications of the technology we develop today. For example, when the cellphone was developed, no one knew that there would be 7.3 billion cellphones in use today. Moreover, he reviewed how connectivity lead to opportunity. In the travel industry, users began to create information and reviewing their travel experiences. had 130 million hotel reviews. 80% of the 18,000 travel agents disappeared almost overnight. He highlighted how complexity also lead to opportunity.
He regailed Duggal Greenhouse with a strategy American Airlines used with reducing the commission rates they were charging Orbitz. As American Airlines titrated commission rates from 10% to 0%, Wayblazer added telephone services and service fees. Mr. Jones closed with a high level mark on how the last 25 years were all about who could build things the cheapest. The next 25 years would be about who can build things the smartest.
Francesco D’Orazio (Face and Pulsar, United Kingdom) continued the session about his thoughts on Watson’s text analytic capabilities in social media. Mr. D’Orazio had three points to make on utilizing Watson’s Alchemy API: 1) A taxonomy could be a specific concept 2) Social media research was not about data mining, rather data surfacing 3) Social media had gone visual. He closed with a graph on the alcoholic drinking habits of fans specific to the type of celebrity. Apparently, Taylor Swift fans definitely do not like to drink Guinness.
Bruce Porter, Ph.D. (University of Texas Austin, USA) was a professor and chair of the University of Texas – Austin Computer Science program. He delineated the difference between artificial intelligence (AI) and intelligence augmentation (IA). AI was the recreation of human cognition. IA was the enhancement of human cognition. Bri Connelly (Cerebri, USA) was the student winner of the IBM Watson Entrepreneur Challenge. She commented on the problem her application was solving – the navigation of social services. Dr. John Kelly III (IBM, USA) closed the opening session with how he was improving clinicians’ abilities to diagnose skin disorders like benign skin lesions and melanoma. After uploading 3000 images of skin lesions to Watson, his team told Watson that 200 images were skin cancer. Watson was then able to identify the texture/coloring of those images to accurately diagnose the respective skin disorders (or lack thereof).

II: New Disruptors, APIs, and Discoveries
Stephen Gold transitioned with comments on how the audience could have visited the Innovation lab upstairs to build their own cognitive application in 30 minutes.
John Adler (IBM Watson, USA) opened the next session with talks on how the world was changing quickly through technology. 81% said mobile was fundamentally changing how they did business. 2.5 billion users were active on Facebook, Twitter, and YouTube. Mr. Adler talked about Watson’s role in concierge services, contact centers, and in-store representative services. The question that he left the audience members with were: 1) How do I personalize the discussion? 2) How do I leverage the user? 3) How do I make the dialogue so human-like that it gave the customer a feeling of assurance that they received the information they needed in a compassionate way?
Jonathan Young (IBM Watson, USA) talked about four companies trying to answer these three questions: GenieMD, AstorTel, Wayblazer, and Personal Bank. He then continued with a conversation of a miniature human-figured robot capable of conversing about mortgage rates. The robot’s name was Mr. Cuddles. Soheil Sadaat (GenieMD, USA), gave insight to Watson’s value in stroke patients. He purported the value of Watson knowing the experiences of patients 1) recovering from a stroke 2) preventing a future stroke 3) exhibiting warning signs of a stroke 4) undergoing treatment plans for a stroke, and 5) Having risk factors for a stroke. Mr. Sadaat closed with a fact on how 10 PubMed articles had been published during the 5 minutes he was on stage. This fact was stated to show how the wealth of knowledge was too overwhelming to keep up with.
Mr. Adler closed the session with an introduction to the Watson Engagement Advisor. The Engagement Advisor resolved issues faster with high customer satisfaction & trust rates. It incorporated Watson’s Q&A API, Watson Dialogue Services, and Watson Curator Collections. He had also hit on the three key phases of “The Cognitive Application Development Journey”: 1) Cognitive Value Assessment 2) Configure and Train 3) Deploy and Manage
Nextly, a second breakout session on the IBM Watson Health started in the developer’s tent. Farhana Alarakhiya (IBM Watson Health, USA) started the panel discussion off with the problem of an aging population with increasing chronic diseases in the United Kingdom. She continued with a comment on enabling three ecosystems: 1) Data 2) Insights 3) Solutions. She also mentioned how important IBM’s partnerships with Apple, Johnson & Johnson, Point of Care, and Geppetto were to IBM Watson’s success. IBM Watson Health’s newest partnership was with Explorys. With 50 million patient records, Explorys would be delivering insights on operation metrics and outcomes metrics.
Amy Frankowski, MD (Mercy Health Select, USA) added to the discussion on the role of accountable care organizations. Her team utilized coordinated care with actionable data to engage patients. Charlie Lougseed (Explorys, USA) chimed in with the value Explorys’ healthcare data analytics services brought to the table. With the largest patient data set in the world, Explorys was enabling healthcare systems that were strained for resources to function. When asked to describe Utopia in a word, Mr.  Lougseed responded with the word, “veracity.” Ashok Rai, MD (Prevea Health, USA) talked about his company’s drive to created patient-centered homes. Moreover, his biggest challenge he faced in terms of extending his “patient-centered home” model to physicians was getting partnering companies interested in the new model.

For the first time in history, IBM and NUI Central co-sponsored a hackathon awarding $25,000 in prize money to those who could develop the best cognitive applications. 2nd place was a cognitive application that utilized the Bluemix platform’s personality insights API. The winner of the hackathon was a cognitive application that connected users to other users with similar interests. All of these cognitive applications were built within 48 hours by about 200 developers from across the world.
In an afternoon breakout session, some of the distinguished IBM staff gave 7 practical lessons on how to build a cognitive application. These lessons were: 1) Gather the right data 2) Design for scale 3) Build a feedback loop 4) Collect user insights 5) Connect with users 6) Turn insights into action 7) Integrate and evolve.

Fish markets, a talking artificial intelligence banking robot, and college students with cognitive application companies – these were just some of the professional components of World of Watson that IBM gave New York City this year. With the advent of utilizing cognitive computing applications to perform tasks ranging from scientific literature review to car salesman training at lightning fast speed, the value of Watson had just come to surface. As a result, clinical outcomes are expected to improve, sales staff services will be more accurate, and customers/patients will feel like they have received optimal service. Unlike any other conference to date, World of Watson had demonstrated the value and successful functionality of consumerized artificial intelligence. The consensus among speakers at World of Watson was clear: cognitive computing science would disrupt many service-based industries.

Yours Truly,
Nicholas Vu, Pharm.D.
Twitter: @nicholasvu