Big Data in Healthcare Flashcards
Definition of Big Data
Data whose scale, distribution, diversity and/or timeliness requires the use of new technical architecture and analysis to enable the insights that unlock new sources of business value
Definition of Real World Data
Data regarding effects of health interventions not collected during RCTs but during routine clinical practice
This includes clinical, economic outcomes, patient reported outcomes
Definition of Real World Evidence
Evidence derived from RWD
Definition of Real World Studies
Non traditional trials looking at effectiveness in routine clinical practice: pragmatic, adaptive trials, drug utilisation studies
Definition of response adaptive randomisation
Procedure that uses past treatment assignments and patient responses to select the probability of future treatment assignments
Definition of adaptive clinical trials
Evaluates an intervention by observing participant outcomes and modifying parameters of a trial according to observations
Definition of reimbursement bias
Data that does not need to be collected into EHRs is not recorded
Definition of software bias
EHR system doesnt allow for certain values to be inputted
What is Big Data
Data whose scale, distribution, diversity and/or timeliness require the use of new technical architectures and analytics to enable insights that unlock new sources of business value
What is Big Data characterised by
4Vs
Volume, zettabytes
Variety, generally not structured, in different forms
Velocity, streaming data that is constantly being generated
Veracity, uncertainty of the quality
How is Big Data used in medicine
Focus on value found in the data and apply it to clinical practice
What are the 2 traditional assumptions of Big Data that don’t apply in medicine
Correlation can replace causality given sufficient no
- We must understand why a=> b before using this knowledge
- If not, might lead to harm
Ignores structure of the population, assumes it is big enough to be valid
-Doesn’t control for bias
What is Real World Data
What are medical examples of RWD
Data regarding effects of health interventions not collected during an RCT
During routine clinical practice
- Clinical, economic and patient reported outcomes
- EHRs, observational studies, patient registries
What is Real World Evidence and Real World Trials
Can derive Real World Evidence from RWD
Real World Studies consist of non traditional trials which investigate effectiveness in routine clinical practice
What are the 4 applications of RWD
- in trials
- observational trials
- demographic studies
- epidemiology
When RCTs cant be used
Assess products and therapies already in use
-observing drug safety and comparing product effectiveness
Recording treatment patterns
-in diverse patient populations to target products and services
Characterise disease and patient populations
-understand epidemiology trends and disease management opportunities
How can pharmaceutical companies use RWD
Value based pricing methods
Gain a better understanding of disease and treatments
Informing product and trial design to address opportunities to gain and defend market access
What are data registries
Collection of info about individuals, usually focused around a specific diagnosis/condition
They are on a voluntary basis
What are claims databases and why are they important in health data collection
What are the 3 pros and 2 cons
Electronic record of claims submitted for the purpose of payment for services in countries without a health service
Pros
- Provides a longitudinal view into a patient’ use of healthcare services
- Economic theory provides assurance of services inclusion
- Typically includes reimbursed patient paid amounts
Cons
- Can lack clinical depth
- Patients are not representative of the population, only those with health insurance included
How would you use social networking data in the future in research
Survey patients on conduct of more patient centric clinical trials
Design of patient reported outcome measures that matter to the patients
Faster way for researchers to conduct observational studies
Mechanism for regulators to collect info on drug related AEs
What are the current views of sharing social media data for care improvement
4 opinions
Generally are happy to share health data
- help doctors improve care
- help research and learn more about their diseases
- help patients like themselves
Many are concerned about health data being used in a detrimental way
Describe the current use of EHR system data in research
Why is primary care data so good
What are the current problems with current EHRs
Data needed extracted from routinely collected clinical data
Primary care data covers cradle => grave
Needs integration with other data sources for treatments outside primary care and context
Hospital EHR systems all vary from department to department
How would you address the downfalls of RCTs with RWD
How would RWD benefit applications for Stage 1 drug trials
May address efficacy, effectiveness gap
By introducing RWD into preauthorisation drug development => improve info available at initial market authorisation
What are the 4 main challenges in clinical trials
Too expensive and difficult to run
Findings apply to the average of a large population, not to targeted subgroups
Or findings are too narrow, population not represented
Morally problematic randomisation
How would you use Big Data in clinical trials
What would the possible effects of this be
Adaptive designs incoorporate rules to adjust trial aspects during enrollment
Use data to pick suitable patients for trials
- generate generalizable estimates of treatment
- understand heterogeneity of treatment effect
How would you use response adaptive randomization in trials
What does it aim to achieve
Randomization procedure that uses past treatment assignments and patient responses to select the probability of future treatment assignments
Aims to reduce the odds of patients being allocated to the poorly faring treatment group
How would you use adaptive clinical trials
What does this aim to achieve
Evaluates an intervention by observing participant outcomes and modifying parameters of a trial according to the observations
This continues throughout the trial
Aims to quickly identify drugs with a therapeutic effect and find the patient population for whom the drug is most appropriate
How would you overcome the slow translation of knowledge
Clinicians can receive the feedback of the data that they entered into the system (results of trials and experiments)
Computer interpretable guidelines that can semi automate the curation of evidence bases
This knowledge can be disseminated to the clinicians
How can EHR data be used to help healthcare providers and researchers
Automatic eligibility checking of patients for trials
Find patients with the appropriate profile for a clinical trial
Automatic insertion of data into electronic forms to save the clinician time
What are the 4 main obstacles to Big Data collection
Restrictive policies on data access
Lack of standard policy of patient data confidentiality
No international standardisation on data collection routes
Expensive and restrictive licenses to data access
Describe the policies on data governance
What are the current problems with data governance
Each data set has a different associated set of info governance regulations
Describe research data governance in the UK
- What type of approval do you need to start a clinical trial
- How can NHS data be used
Clinical trials need ethical approval from National Research Committee
NHS data collected for clinical/admin purposes can be used without consent but not always for research even though the usage of the data is the same in observational studies
What are the 3 main problems with routine data quality
Data errors
-use of abbreviations and terminology not standardised
Reimbursement bias
-Data that doesnt need to be collected is not recorded
Software bias
-System doesn’t allow certain values to be inputted
