Precision Health and Imaging Analytics

Transcription

1 17 th Annual International Healthcare Summit Kelowna, BC Precision Health and Imaging Analytics Charlotte Hovet, MD, MMM Medical Director, Healthcare Solutions & Consulting June 26, NTT DATA, Inc. All rights reserved.

2 Precision Health Predicting and preventing diseases before they manifest by tapping into health data to provide targeted, predictive and personalized care NTT DATA, Inc. All rights reserved. 2

3 Impact of Data-Driven Care Management 2017 NTT DATA, Inc. All rights reserved. 3

4 Evolution of Healthcare Delivery As healthcare delivery evolves towards collaborative care models, the ability to share data and use it to improve decision making is a key transformative milestone Phase 1 Phase 2 Phase 3 share data and use it to improve decision making will be a key transformative milestone Move and exchange data EMR Lab/eRx Hospital Physician Payer Manage patient health Capture and digitize records Electronic medical record EMR Patient health management Interoperability Information driven decision making Population health EMR EMR BI & analytics Analyze and manage data 2017 NTT DATA, Inc. All rights reserved. 4

5 Imaging Analytics Imaging analytics uses machine vision and machine learning to analyze vast amounts of data and automatically detect patterns and features NTT DATA, Inc. All rights reserved. 5

6 What is Machine Learning? Classic ML Using optimized functions or algorithms to extract insights from data Director of Deep Engineering, learningintel Health Using massive labeled data sets to train deep (neural) graphs that can make inferences about new data Training Data * Algorithms Random Forest Support Vector Machines Regression Naïve Bayes Hidden Markov K-Means Clustering Ensemble Methods More Inference, Clustering, or Classification Untrained Step 1: Training Hours to Days in Cloud CNN, RNN, RBM.. New Data Trained Step 2: Inference Real-Time at Edge/Cloud New Data * Use massive labeled dataset (e.g. 10M tagged images) to iteratively adjust weighting of neural network connections Form inference about new input data (e.g. a photo) using trained neural network *Note: not all classic machine learning functions require training Source: Prashant Shah, Director of Engineering, Intel Health and Life Sciences 2017 NTT DATA, Inc. All rights reserved. 6

7 Advances in Machine Vision and Machine Learning in Medical Imaging 3D visualization is an important tool but it is not machine vision Machine vision is the ability to automate and identify anatomy and measure its characteristics Machine learning takes one or more measurements and relates them to disease states based on learned patterns Proper application of these technologies create actionable insights impacting patient care» Not just more data points but recommendations on clinical pathways and patient risk 2017 NTT DATA, Inc. All rights reserved. 7

8 Imaging Across the Continuum of Care Imaging is part of all phases in healthcare Screening Diagnosis Treatment Imaging is accessed by multiple participants Nurses Primary care physicians Interventionists (e.g. surgeons, cardiologists) Specialists (e.g. oncologists) Patients Collaboration is changing more than just sharing results across multiple users accessing multiple clinical applications potentially across multiple facilities Clinical images, scanned documents, HL7 reports, video, still images, waveforms 2017 NTT DATA, Inc. All rights reserved. 8

9 The Value in Imaging Insights Picture This A metaphor for clinical imaging look at this picture and identify who is from out of town 2017 NTT DATA, Inc. All rights reserved. 9

10 The Value in Imaging Insights A metaphor for clinical imaging incidental findings Beyond identifying who is from out of town, tools can help show when, where, and what was bought as incidental findings 2017 NTT DATA, Inc. All rights reserved. 10

11 Imaging Analytics: Uncovering Hidden Risks Diagnostic imaging uniquely positioned to identify incidental findings Why imaging? Why incidental findings? A standard of care in diagnostic process Lowest cost of identifying disease Event to baseline patients health beyond the prescribed reason for exam Opportunity for comparison with previous images to track change Identify patients at risk profiles High value event to reduce cost of care and increase safety (incident avoidance and dose reporting) Advances in imaging technology have made incidental findings available Improve patient outcomes Source: Imaging Performance Partnership interviews & analysis 2017 NTT DATA, Inc. All rights reserved. 11

12 Imaging Analytics: A Growing Library of Tools A platform* for leveraging diagnostic imaging procedures A secure connection Imaging studies filtered by institutional criteria and sent for screening analysis Results range from incidental findings to consistent quantitative documentation Results can be sent to the VNA, PACS, EHR, and/or care management solution Growing Library of Tools for Disease Detection Lung Elasticity Brain Mapping** Emphysema Breast** Pulmonary hypertension** Fatty liver Coronary calcium scoring Osteoporosis * Platform Powered by NTT Data, Inc./Zebra Medical Vision, LLC ** Algorithm not available for sale in the U.S., Pending FDA 2017 NTT DATA, Inc. All rights reserved. 12

13 Integrating into Diagnostic Imaging Automated quantification for aiding diagnostic reading Patient Presents for Primary Diagnostic Study Primary and Incidental Finding Report Published Study Completed and Diagnostic Images Read Automated Analytics Reports Measurements from Anatomic Specific Algorithms Radiologist Reports with Prepopulated Detail Template Primary Care Team Foreseen Value: Easier detection of asymptomatic disease from existing studies (incidental finding) Interpretation and productivity enhancement for radiology, cardiology and future areas of imaging 2017 NTT DATA, Inc. All rights reserved. 13

14 Alternate Integration of Imaging Analytics Options for automated initiation of screening H i g h R i s k L o w L o w Hidde n opport unity Rel ati vel y he alt hy Cost (PMPM) Cri tic al Hi gh util ize rs H i g h Population health consumption stratification using ordered studies Patient eligibility and imaging study type Automated Imaging Analytics Primary care selection of screening with principle diagnostic request Patient Disease Screening Result(s) Primary Care Team 2017 NTT DATA, Inc. All rights reserved. 14

15 The Benefits of Imaging Analytics Precision health depends on the use of all available health data, and full use of existing diagnostic images requires a coordinated partnership between the radiologist, the primary care physician/team and the patient. Provider Clinical data sets Analytics Platform Clinical insights 2017 NTT DATA, Inc. All rights reserved. 15

16 The Goal of Healthcare is to Optimize Health and Well-being Health is Wealth 2017 NTT DATA, Inc. All rights reserved. 16

17 2017 NTT DATA, Inc. All rights reserved.