Huisarts van de Toekomst Martijn G H van

FLU (influenza) TRENDS http: //www. google. com/flutrends

Decision Support Patient communication Multidisciplinary Tumor Boards

Data overflow • Medical information doubles every 5 years. By 2020 it is expected

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Decision Support - examples “Rule-based” “Machine-learning”

Decision Support - examples “Rule-based” • Stand-alone system • Local • Available as App

Oncoguide Future perspectives: § Continuous updates of guideline § Clinical Trial Matching § Regional

IBM Watson for Oncology 2012 IBM partners with Memorial Sloan Kettering to train their

Watson for Oncology Advisor Clinical Trial Matching Genomics Watson for Oncology - Platform Imaging

Challenges • No peer reviewed papers available about efficacy or (cost)effectiveness • “Black box”

Framework / Ecosystem 1. 2. 3. 4. 5. Quality Assurance Technical Assurance Clinical Implications

Ethical considerations What you say on Twitter may be viewed all around the world

Tweets with “Crohn” N=6, 084 Language = English and Dutch N=5, 421 Original tweets

Results – Orginal tweets ‘I haven't eaten food in 2 days and no appetite

Results – Patient’ tweets @genag 515 Wish I was full of energy and joys.

Comparison with focus groups • Computerized Twitter search “Heartburn” • Collection period: 1 week

Saturation Curve curve to hit all domains once 120 100 Percent saturation 80 60

Saturation Curve curve to hit all domains once 120 100 Percent saturation 80 NNTweet

Wearables http: //www. watch-society. com/

AGIS - “Ab. Stats” J Gastrointest Surg 2014; 18: 1795 -1803

Wearables in Oncology Fitbit HR: • to measure Performance Status • to predict adverse

Slides: 59

Download presentation

Huisarts van de Toekomst Martijn G. H. van Oijen, Ph. D Associate Professor Dept. Medical Oncology

Disclaimer No financial disclosures

Meet your new colleagues

Meet Dr. Bing

Meet Dr. Google

BMJ 2014; 339: g 7392

57. 7% correct BMJ 2014; 339: g 7392

BMJ 2014; 339: g 7392

FLU (influenza) TRENDS http: //www. google. com/flutrends

Google Flu Trends

Google Dengue Trends

Meet Dr. Watson

Decision Support Patient communication Multidisciplinary Tumor Boards

Data overflow • Medical information doubles every 5 years. By 2020 it is expected to double every quarter • 80% of the healthcare professionals spends at most 5 hrs/month to keep abreast of his/her domain • 80% of the information is unstructured

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 “colorectal cancer” Number of publications 16000 14000 12000 10000 8000 6000 4000 2000 0

Decision Support - examples “Rule-based” “Machine-learning”

Decision Support - examples “Rule-based” • Stand-alone system • Local • Available as App • Based on guidelines • No integration with EHR • Free of charge “Machine-learning” • Stand-alone system • Cloud solution • Can be used on Tablet • Based on “all available knowledge” • Minimal integration with EHR • Annual license

Oncoguide

Oncoguide Future perspectives: § Continuous updates of guideline § Clinical Trial Matching § Regional Referral Options

1997 - Deep. Blue

2011 - Jeopardy

One size fits all

IBM Watson for Oncology 2012 IBM partners with Memorial Sloan Kettering to train their supercomputer Watson 2014 First pilot study results submitted to ASCO 2015 Poster presentation at ASCO

Results

Watson for Oncology Advisor Clinical Trial Matching Genomics Watson for Oncology - Platform Imaging

Challenges • No peer reviewed papers available about efficacy or (cost)effectiveness • “Black box” • Legal and ethical issues • Automated data extraction • Continuous evaluation of ‘updates’ • Reimbursement

Framework / Ecosystem 1. 2. 3. 4. 5. Quality Assurance Technical Assurance Clinical Implications Practical Evaluation Reimbursement Schippers signs “Health Deal” – 18 -6 -2016

Chef Watson www. ibmchefwatson. com

Meet Dr. Twitter

Created by Eric Fischer

Ethical considerations What you say on Twitter may be viewed all around the world instantly. I Agree

Tweets with “Crohn” N=6, 084 Language = English and Dutch N=5, 421 Original tweets N=2, 236 (41%) Patients N=987 (44%) Friends or Relatives N=484 (22%) Re-tweets N=3, 185 (59%) Non-Profit or Foundations N=432 (19%) Healthcare Professionals N=167 (7%) Pharma/ Biotech N=137 (6%) Unclassified N=29(1%)

Results – Orginal tweets ‘I haven't eaten food in 2 days and no appetite whatsoever. I’m used to this with #Crohn's’ ‘Found out my friend is in the hospital with Crohn's and has to have a bowel resection. I never knew : (‘ Van Oijen et al. (DDW 2012)

Results – Patient’ tweets @genag 515 Wish I was full of energy and joys. . But on Social Security right now due to crohn's, I hope I’m going back to work very soon ‘Seizure caused by low calcium, Vit. D & magnesium. But I don't have any direct Crohn's symptoms now. ’ Van Oijen et al. (DDW 2012)

Comparison with focus groups • Computerized Twitter search “Heartburn” • Collection period: 1 week • Manual categorization of tweets • Comparison with focus groups from GI-PROMIS initiative

Saturation Curve curve to hit all domains once 120 100 Percent saturation 80 60 40 20 0 0 30 60 90 120 150 180 210 240 Number of Tweets 270 300 330 360 390 420 Baek et al. (DDW 2013)

Saturation Curve curve to hit all domains once 120 100 Percent saturation 80 NNTweet = 388 60 40 20 0 0 30 60 90 120 150 180 210 240 Number of Tweets 270 300 330 360 390 420 Baek et al. (DDW 2013)

Meet Dr. Fitbit

Admitted @home

James

Wearables http: //www. watch-society. com/

AGIS - “Ab. Stats” J Gastrointest Surg 2014; 18: 1795 -1803

Wearables in Oncology Fitbit HR: • to measure Performance Status • to predict adverse events • to study quality of life Apple i. Phone: • to test for phase I study eligibility

Huisarts van de Toekomst Martijn G. H. van Oijen, Ph. D Associate Professor Dept. Medical Oncology