User Needs and Requirements Analysis for Big Data Healthcare - - PDF document
User Needs and Requirements Analysis for Big Data Healthcare Applications Sonja Zillner, Siemens AG In collaboration with: Nelia Lasierra, Werner Faix, and Sabrina Neururer MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062 #
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Sonja Zillner, Siemens AG In collaboration with: Nelia Lasierra, Werner Faix, and Sabrina Neururer
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Europe needs a clear strategy for leveraging Big Data Economy in Europe Work at technical, business and policy levels, shaping the future through the positioning of Big Data in Horizon 2020. Bringing the necessary stakeholders into a sustainable industry-led initiative, which will greatly contribute to enhance the EU competitiveness taking full advantage of Big Data technologies. Objectives Trigger Type of project: Coordination & Support Action Project start date: September 2012 Duration: 26 months Call: FP7-ICT-2011-8 Budget: 3,038 M€ Consortium: 11 partners Facts
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Definition of Big Data in Healthcare Industry
healthcare business intelligence, health data analytics and health data management application
handling and analysis of complex and large healthcare data by relying on
Characteristics of Health data
for medical images and NGS is immature and in development
under the label „Advanced Health Data Analytics“
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
development, payors, medical product providers
1
2
3
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
interventions in order to increase efficiency and quality of clinical care services.
heterogeneous health data sets as well as advanced analytics
them accordingly, e.g. analyzing medical procedures to find performance opportunities, such as improved clinical processes, fine-tuning and adaptation of clinical guidelines
clinical, financial, administrative as well as other related health data in order to discover new valuable knowledge, for instance to identify trends, predict outcomes or influence patient care, drug development, or therapy choices, e.g.
cost structure of care settings, etc. By using nation-wide disease registries, i.e. databases covering secondary data related to patients with a specific diagnosis, condition or procedure.
engagement of patients in their healthcare process.
1 2 3 4 5 6
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Improved Efficiency of Care 1
resource utilization
reporting Improved Quality of Care 1
1= Frost & Sullivan “U.S. Hospital Health Data Analytics Market (2012) ”
Real Impact
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Clinical Data
hospitals, care centers, physicians, etc.)
within the classical hospital information systems or EHR, such as medical records, medical images, lab results, genetic data, etc. Claims, Cost & Administrative Data
reimbursement issues, such as utilization of care, cost estimates, claims, etc. Pharmaceutical & R&D Data
companies, research labs/academia, government
clinical studies, population and disease data, etc. Patient Behaviour & Sentiment Data
producer
information related to the patient behaviours and preferences Health data on the web
websites such as PatientLikeMe, Linked Open Data, etc.
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
expenditure and at the same time help to increase the quality of care settings
diagnosis, right treatment to right patient, rapid cycle time of treatments, fewer complications, fewer mistakes, slower disease progression, etc.
Value-based healthcare is becoming focus of many healthcare reforms Principles 1 Example
1= Porter and Olmsted Teisberg. “Redefining German Health Care”, 2006
means to track and analyze treatment performance of patients and patient populations Big Data Technology....
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Drivers
US Healthcare Reforms HITECH & PPACA Healthcare Reforms HITECH & PPACA
Constraints
affects analytics usage
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
High Investment Long-term investments require conjoint
engagement of several partners
Data Digitalization
documented (lack of time) with low quality
Semantic Annotation transform unstructured data into
structured format
Data Sharing Overcome data silos and
inflexible interfaces
Business Cases Undiscovered und unclaimed
potential business values
Value-based system incentives Current incentives enforce “high
number” instead of “high quality” of care services
1 1
Not-Technology-related
1 2 3 2 3
Data Security Legal processes for data sharing &
communication are needed
Regulation & Technology Technology-related
2
Data Quality Reliable insights for health-related
decisions require high data quality
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
Technology-wise = Evolution
technologies, machine learning, scalable data storage, etc.) is ready to be used
data paradigm
data anonymization, understand analytic needs) Business-wise = Revolution
boundaries relies on effective cooperation of multiple stakeholder with diverging interests => existing industrial business processes will change fundamentally, new players & business models will emerge
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
healthcare cost, increased need for healthcare coverage and shifts in provider reimbursement trends trigger the demand for big data technology.
more efforts are needed
analytics technologies, are theoretically in place
data technology in healthcare
healthcare system which hinder collaboration and, thus, data sharing and exchange
collaboration to enhance the treatment patient of the patient, and thus will significantly foster the need for big data applications
1 2 3 4 5 6
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ know-why ▶ Topics to be addressed: market, customer, competitor, environment, trends, threads, constraints, driver
Market Layer (SF)
▶ know-how ▶ Topics to be addressed: technologies, competences, knowledge, supplier, infrastructure, standards
Technology Layer (TG)
▶ know-what ▶ In BIG: application/ use case scenarios ▶ Example of Health SF ▶ Scenario1: Comparative Effectiveness Research ▶ Scenario2: Next Generation of CDS ▶ Scenario3: Patient Profiling for Risk Identification……
Product Application Layer (Cross-Sectorial Work)
▶ Includes the description ▶ Technological Requirements ▶ Market Impact
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Sector-specific data management technologies that need to be in place before any big data scenario can be realized ▶ Ensure that the relevant data of the sector is available (e.g. EHR, IED) ▶ Driven by user/need driven (sector) analysis ▶ Focus on domain-specific requirements of data management and related research questions
Enabling Technologies
▶ Large Gap between needs mentioned by the stakeholder and users of the (health) sector and the technological opportunities envisioned by the technical groups ▶ Technical requirements mentioned within sectorial interviews mainly relate to efficient data management approaches ▶ Technical opportunities mentioned by the technical groups highlight opportunities within a world
▶ Challenge: How to align the two perspectives? ▶ First Step “Focus on big data readiness”: Any technological requirements, such as efficient data management, need to be addressed /solved before big data capabilities can be implemented (enabling technologies) ▶ Second Step: “Elaborate big data opportunities”: Develop transitional scenarios that could be realized assuming that the sector has achieved big data readiness (big data technologies), scenarios should generate value.
Observations & Learnings from 1st sector and technical investigations
▶ advanced/big IT capabilities that help to improve healthcare delivery ▶ Data-driven: Investigate in public available health data sources as basis for use case brainstorming ▶ Technology-Driven: Investigate to which extent applications /technologies from other domains can be transferred to the healthcare sector
Big data opportunities
We need to distinguish between
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Compare the clinical and financial effectiveness of interventions in order to increase efficiency and quality of clinical care services. ▶ Large datasets encompassing clinical data (information about patient characteristics), financial data and administrative (treatments and services accomplished) are critically analyzed in order to identify the most effective treatments that work best for particular patients.
Description Stakeholder’s benefit Example Applications
Several public-funded research agencies, such as the National Institute for Health and Care Excellence (UK), Institute for Quality and Efficiency in Healthcare (Germany), the Common Drug Review (Canada) or the Australian’s Pharmaceutical Benefits Scheme, started to run Comparative Effectiveness Research programs
Prerequisites
▶ Data Digitalization and Data Integration of health data from various domains ▶ High data quality with broad coverage. ▶ Avoidance of biased Data Sets: ▶ Clinicians could receive recommendation about the most clinical effective treatment alternative for a particular patient. ▶ Hospital operator could receive a recommendation about the most financial effective treatment alternative for a particular patient. ▶ Payors could use the discovered knowledge about the most effective treatment to align their reimbursement strategy. ▶ Patients could benefit by receiving the most effective treatment in accordance to their particular health conditions.
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Clinical decision support (CDS) applications aim to enhance the efficiency and quality of care
professionals in their decision making process by enabling context-dependent information access, by providing pre-diagnose information or by validating and correcting of data provided. ▶ Thus, those systems support clinicians in informed decision making, which again helps to reduce treatment errors as well as helps to improve efficiency. ▶ By relying on big data technology, future clinical decisions support applications will become become substantially more intelligent. substantially more intelligent.
Description Example Applications Prerequisites
▶ Trust and Confidence are crucial that CDS systems will be accepted. ▶ As clinicians will only rely on CDS systems, if it is guaranteed that all relevant data sources are integrated, the aspect of comprehensive data integration in high data quality is an important prerequisite. ▶ Pre-diagnosis of medical images, ▶ Treatment recommendation reflecting existing medical guidelines. ▶ etc.
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Clinical Operation Intelligence aims to identify waste in clinical processes in order to optimize them accordingly. ▶ By analyzing medical procedures, performance
tuning and adaptation of clinical guidelines, can be realized.
Description Stakeholder’s benefit Example Applications Prerequisites
▶ Data Security and Privacy Requirements: seamless data access & a common legal framework regulating the use of patient data. ▶ Engagement of Clinicians: Adaptations regarding clinical guidelines need to be initiated and approved by healthcare professionals ▶ Healthcare professionals gain further insights into the effectiveness of treatment decisions and processes and can adapt their decisions accordingly. ▶ Patients are informed about the effectiveness of treatments and can select those treatments that offer best value for them.
Example Use Cases
▶ Publishing of cost, quality and performance data
competition that in consequence will drive performance improvements.
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Secondary usage of health data is defined as the aggregation, analysis and concise presentation of clinical, financial, administrative as well as other related health data ▶ in order to discover new valuable knowledge, for instance to identify trends, predict outcomes or influence patient care, drug development, or therapy choices.
Description Example Use Cases Prerequisites
▶ Integrated data in high quality: the value of data analysis depends on the integration of comprehensive and complete data as well as on the quality of input data ▶ Business case: any successful implementation requires a clear business case. ▶ Privacy and security of data: a common legal framework specifying data access control & policies of data usage ▶ Standards, e.g. ICD-10 or HL7, are needed to establish a common semantic (re-) used data items ▶ Example 1: Identification of patients with rare disease Big data technology is used to identify (early detection) of patients with rare diseases. ▶ Example 2: Patient recruiting and profiling suitable for conducting clinical studies. ▶ Example 3: Forecast of clinical process values Comprehensive health data sets are analyzed in
regarding relevant clinical benchmarks ▶ Example 4: Health knowledge broker Health- related data is analyzed to develop commercialization plans or portfolio strategies for third party companies.
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ Public health analytics applications rely on the comprehensive disease management ▶
▶ severe (e.g. cancer) diseases ▶ that allow to aggregate and analyse treatment and outcome data which again can be used to reduce complications, slow diseases’ progression, as well as improve outcome, etc.
Description Stakeholder’s benefit Example Applications Prerequisites
▶ Clinical Engagement, i.e. active engagement, clear responsibility for data collection and interpretation, by the clinical community ▶ National Infrastructure, i.e. common standards, shared IT platform and common legal framework defining the privacy & security requirements for tracking diagnosis, treatments, outcomes on patient level ▶ High-Quality Data achieved through systematic analysis of outcome data of a patient population ▶ System Incentives that rely on the active dissemination and usage of outcome data ▶ Payors: As of today, payors lack the data infrastructure required to track diagnoses, treatments, outcomes, and costs on the patient level and thus are not capable to identify best- practice treatments ▶ Government: can reduce healthcare cost & improved quality of care ▶ Patients: can access improved treatments according to best-practice knowhow ▶ Clinicians: usage of best-practice recommendation and informed decision making in case of rare diseases
Success Story: Sweden
▶ Since 1970, Sweden established 90 registries that cover today 90 % of all Swedish patient data with selected characteristics (some cover even longitudinal data). A recent study showed that Sweden has best health-care
‹#› MIE 2014 in Istanbul: 01-09-2014 / Zillner BIG 318062
▶ The idea is to establish a platform/patient portal that fosters the active patient engagement in the context of patients’ health care processes. ▶ The patient platform offers smart phone apps and devices to its members/patients to monitor health-related parameters, such as activity, diet, sleep or weight. ▶ The underlying assumption is that patients who are able to continuously monitor their health-related data are encouraged to improve their life style as well as improve their own care conditions.
Description Stakeholder’s benefit Example Applications Prerequisites
▶ Business case: a convincing business case describing the business value and model needs to be investigated. ▶ Commitment of stakeholder, such as patients, clinicians, and payors. ▶ Patients: The increased patient engagement through actively producing and providing health data might improve overall wellness and health conditions. ▶ Payors/Government: Reduced healthcare cost through preventive care. ▶ Clinicians: Informed decision making through access to heterogeneous data sources, such as biometric, device or clinical data. ▶ Development of predictive models that allow evaluating and predicting the successful patient behaviour in a particular health program can help ▶ to replicate supporting influence factors or ▶ to identify reasons why some patient gave up a program, etc.