Big Data in Healthcare Raj Dalal Principal CCC

Big Data in Healthcare Raj Dalal Principal – CC&C Solutions Founder - Big. Insights

. • A Consulting company, established in 2002 with primary focus on Enterprise and Solution Architectures. • Providing thought leadership in as a Gold member • Operating globally through dedicated professional teams • Drawing on international best practices from TOGAF ® & Other standards • Providing Professional Services to numerous enterprises with a primary focus on: – Health services – Public Sector – Financial Services & Insurance – Supply Chain & Distribution – Education – IT service providers

Some of our worldwide customers. Fortune 500 Corporations, Leading Government Agencies, Global Service Providers, Universities

New era of accelerated transformation

Big Data

Primary data pools are at the heart of the big data revolution in healthcare Source: Mc. Kinsey Global Institute

Big Data: Foundation for evolution of healthcare Source: Hewlett Packard

Empowering the Individual/Patient

Personalised consumer healthcare Source: Hewlett Packard

Big Data is changing the paradigm: New value pathways in healthcare Source: Mc. Kinsey Global Institute

Application of early successes at scale could reduce US healthcare costs by $300 -450 Billion

Big Data Use Cases Source: Flutura

Key word Analysis - Keyword mining of doctor’s/Lab transcripts using text mining and co-relations to patient outcomes Source: Flutura

Source: Flutura

Telemedicine Analytics Source: Flutura

Apriori sequence analysis to define new clinical pathways Source: Flutura

Source: Flutura

Predictive Analytics • Improve patient experiences through continuum of care by examining referral patterns across providers • Track spread of infection inside hospitals through wellplaced sensors; predict risk of outbreaks through monitoring of multiple data sources • Clinical revolution: predictive analytics for diagnosis, risk assessment, prognosis, therapeutic management • Learn best ways of contacting patients to ensure that they follow treatments, nutrition guidelines, and can recognize symptoms

Healthcare providers must develop a range of big data analytics capabilities

Case Study US Depart of Veterans Affairs

Overview of Project • US Veterans are committing suicide at an alarming rate • Some disturbing statistics: – 22 US veterans per day commit suicide – Suicide rates are roughly double those of general US population – 21% of overall suicides in USA are veterans – 12, 500 calls per month to veterans crisis line (30% are related of suicide) • Durkeim project aims to proactively predict negative events such as suicide in the US veterans population and prevent it.

Outcomes • Dedicated to applied research on predictive suicide risk: – Multidisciplinary team of psychiatrists, US Veterans Administration and AI (Machine learning) • Phase 1: – Developed linguistic-driven prediction models to estimate the risk of suicide. (from VA medical records). – Generated Datasets of single keywords and multiword phrases and constructed prediction models using machine learning algorithms

Outcomes – Phase 2 • Sucidality prediction at scale • Expand the data available for VA physicians with real time social media feeds from 100, 000+ US Veterans to medical database • Clinicians get a dashboard to monitor risks based on this realtime data • Can provide real-time triage upon detection of critical event

Durkheim Project (Phase 2) - Proactive Opt-In Signup • Enabled privacy controls for user • Both ‘opt-in’ and ‘opt-out’ options are supported • Users are informed of the social networks that are enabled Realtime monitoring of Social Media: • Facebook post and status Updates • Twitter • Google+ (coming soon)

Clinician dashboard • Patients are listed by with overall risk rating/statistics and clinical notes • Patients flagged for risk ie Green (Low) to Red (High) • Status updated every 1 minute Drill Down: - Suicide Ideation Rating - Time Series/Trend (Mental health ticker) -- Risk against other patients

Clinician dashboard Clinician notes: Combined analysis of patients self -reported text as well as observations Source content: Can view the social media updates

Technology Medical Database residing behind the medical firewall at Geisel School of Medicine where participant data is stored for analysis by medical professionals according to Human Subject Study/ HIPAA privacy rules Information technology: • Hadoop technology use to store potentially massive datasets • Attivio used for unstructured search • Cloudera Search and Cloudera Impala being introduced into their machine learning framework to simplify the environment and reduce data movement Statistical tools • Bayesian counters (B-counts) used for on-line near realtime model building and prediction. Underlying technology for B-counts is HBase. • Initial prediction algorithm is Naïve Bayes (NB). The framework is currently being extended to incorporate Nearest Neighbours (NN) and Bayesian Network (BN) learning algorithms.

Benefits/Metrics • Phase 1 - Data model’s predictive accuracy was statistically significant (at least 65% accurate) • Proactive platform for refining predictive models and suicide prevention at scale

Summary Focus Area US Veterans committing suicide Outcomes Reduction in number of Veterans committing suicide Sources Past Veterans History Social media monitoring Metrics Data model’s predictive accuracy was statistically significant (at least 65% accurate)

New Style of IT in healthcare Shaped by cloud, Big Data, security and mobility

EA in Healthcare transformation National Health vision, strategy and blue print for Healthy Nation Users portal- 720 degree view Business Layer Traditional | Collaborative | Outcome and Data Driven Data Layer Structured | Un-Structured | Real-Time | Bigdata Application Layer Administration | Drug control | Tele Medicine | Real time and Continuous Monitoring (Internet Of Things) (Services - based Models) Technology Layer On Premises -> Service-based Models (Cloud) | Hadoop | No. SQL

CC&C Solutions Focus • Big Data integral in Enterprise Architecture transformation • Leadership in Open Group – standardisation (Platform 3. 0) • EA / TOGAF / Archimate training and consultation • Experiences in Health. Care domain – F 100 customer (Johnson & Johnson) – E-health in various Asian countries and NSW Health

Thank You / Questions Contact : Raj Dalal / raj@ccandcsolutions. com / @Big. Insights