Lecture 1 - Medical Data PDF

Lecture 1 - Medical Data Learning Objectives Medical data - types and uses Data vs Information vs Knowledge Case studies What is Medical Data? Gathering and interpreting their meaning are central to the health-care process Virtually all medical-care activities involve gathering, analysing or using d...

Lecture 1 - Medical Data Learning Objectives Medical data - types and uses Data vs Information vs Knowledge Case studies What is Medical Data? Gathering and interpreting their meaning are central to the health-care process Virtually all medical-care activities involve gathering, analysing or using data. Provides a basis for categorising the problems a patient may be having Identify subgroups within a population of patients Assist the decision of treatments Datum: single observation of a patient Temperature reading Red blood cell count Past history of Covid-19 Blood-pressure reading Example: Blood Pressure A reading of 120/80 can be recorded as a single data point Or as two pieces of information Systolic pressure = 120 mm Hg Lecture 1 - Medical Data 1 Diastolic pressure = 80 mm Hg Modifiers: Additional Information Taken from the leg or arm? Lying or standing? Following a prolonged period of exercise All these modifiers can alter the result you record from a blood pressure reading Data may not just be a number but may have to include documentation showing HOW the data was recorded. Types of Medical Data Includes a broad range of data types: Narrative Textual Data Numerical Measurements Recorded Signals Images Homework: Compare the different types of medical data reported in terms of: Ease of acquisition Information density Cost Datum size Invasiveness Privacy Uses of Medical Data Lecture 1 - Medical Data 2 Individual (patient-specific) Support the care of the patient Society (population-based) Aggregation and analysis of data regarding populations of individuals Enables doctors to learn from experience Patient history Symptoms reported Physical Laboratory/Exam results Support communication amongst GPs Anticipate future health problems Monitoring for excess of weight, high blood pressure, elevated cholesterol levels Support clinical research Identify deviations from expected trends (deviations from desirable standards) Baby growth chart BMI charts Overarching Goals and Potential Personalised Medicine Precision Medicine Challenges in using Medical Data Sheer diversity and complexity of data from disparate sources Lack of standardisation - large number of proprietary and legacy systems that do not talk to each other Lecture 1 - Medical Data 3 Missing datapoints Concerns about privacy and exploitation (more about this later) Complex regulatory landscape In the UK: NICE, MHRA, DoH, RCP, (EU) Data vs Information vs Knowledge Data are the physical entities at the lowest abstraction level (value) e.g. generated by a patient (patient data) or biological process. Information is derived from data by interpretation Information is the process of interpreting the data Knowledge is obtained by inductive reasoning with previously interpreted data collected from similar patients or processes. Based on previous data I believe this will be true about this patient who I know nothing about but have measured this data from Example A patient with blood has blood pressure of 180/110 A patient with the blood pressure of 180/110 is a datum as it is the report the patient has hard a heart attack When researchers pool and analyse such data to determine that 180/110 corresponds to a high blood pressure (information) They identify that patients with high blood pressure are more likely to have heart attacks than patients with normal or low blood pressure (knowledge) Going from data to information to knowledge is the problem technology companies are trying to solve. It is difficult because: There is still a lot of data on paper files which is static Lecture 1 - Medical Data 4 Computational systems offer immense power and opportunity to act based on medical data in ways that were unimaginable in the past. To exploit this, computational systems need standardised vocabulary and nomenclature In data storage, retrieval, analytics and outputs Imprecision and the lack of a standardised vocabulary are problematic when we want to Aggregate over the data Analyse trends over time Biomedical Data Sources Every growing libraries of life science data, collected from scientific experiments and computational analyses Are accumulating data across a wide range of scales: from the gene/cell level through to the human and societal level. Lecture 1 - Medical Data 5