HLTH 230 Final: Final Exam w/o TED Talk (Unit 4-7)

Unit 4 Lecture 1- Conventional Medical Decision Making Evidence Based Medicine: Conscientious, explicit, and judicious use of the best current evidence in making decisions about the care of patients - Evidence can be retrieved from published articles, papers, and other scholarly articles. Case Report: Description of individual clinical cases - Article that describes and interprets individual case in a form of a written story - Describes: o Unqiue cases that don’t have diseases or symptoms o Importatnt variation of a disease/condition o Patient has 1 or 2 unexpected diseases Observational studies: 1. Case-control study (Retrospective): a. Calculates odds ratios and population risk with confidence intervals b. Compares patients who have a disease/outcome of interest (Cases) w/ patients who don’t have the disease/outcome (Controls). Looks back w/ consideration of past events (retrospectively) to compare how often the exposure to a risk factor is present in each group. Therefore, determining the relationship b/w the risk factor and the disease. c. Pros: i. Good for studying rare conditions ii. Not a lot of time needed to do study b/c disease has struck iii. Can look at multiple risk factors at once d. Cons: i. Problems with data quality, rely more on memory and people w/ condition ii. Not good to evaluate diagnostic test iii. Difficult to find a suitable control group 2. Cohort Study (Retrospective/Historical): a. One or more samples (cohorts) are followed prospectively and subsequently a status evaluation is conducted to determine which participant has exposure characteristic (risk factors) in respect to the disease in question. b. Pros: i. People of same cohort therefore limiting confouding variables ii. Easier and cheaper thanr Randomized control trial c. Cons: i. Not randomization ii. Blinding is difficult iii. Outcome of interest could take time to occur Prospective Cohort Study: The distinguishing feature of a prospective cohort study is that at the time that the investigators begin enrolling subjects and collecting baseline exposure information, none of the subjects has developed any of the outcomes of interest. Retrospective Cohort Study: The distinguishing feature of a retrospective cohort study is that the investigators conceive the study and begin identifying and enrolling subjects after outcomes have already occurred. Confounding variables: When examining the association between an exposure and an outcome, and risk factors may distort the magnitude of association if they are unequally distributed between the groups being compared. Eg. studying the association between heavy smoking and lung cancer, the two exposure groups (heavy smokers & non-smokers) might differ substantially in age distribution, and age is an independent risk factor for lung cancer. If smokers were older than non-smokers, then the unequal age distribution would exaggerate the strength of association between smoking and lung cancer. Randomization minimizes confounding Randomized Control Trials: A study design that randomly assigns participants into an experimental group or a control group. As the study is conducted, the only expected difference between the control and experimental groups in a randomized controlled trial (RCT) is the outcome variable being studied. Only difference b/w the study and control group should be exposure OBSERVATIONAL STUDIES CAN NEVER COMPLETELY REMOVE CONFOUNDING AND OTHER BIASES Pros: - No population Bias - Easy to blind - Can be analyzed with known stats tools - Individuals are clearly identified Cons: - Expensive in terms of money ($12M) and time - Participants won’t necessarily represent the population as a whole - Doesn’t reveal causation - Time lag b/w study and publication - Strict inclusion and exclusion criteria - Ethical Issues Blinding/Masking: When the groups have been randomly selected from a population do not know whether they are in the control group or the experimental group - Single-Blind: Patients don’t know which group they are in; Only clinicians and researchers do - Double Blind: Patients and clinicians don’t know which patient is in which group; Only the researchers do - Triple-Blind: Patients, clinicians, and researchers don’t know which patient is in which group Evidence Synthesis: Systematic Review: - Document written by a panel providing comprehensive review of all relevant studies on A particular health-related topic/question - Created AFTER reviewing and combing all the information from both published and unpublished studies (focuses on Clinical trials of similar treatment) and then summarizes it - Evidence based source and more reliable than individual studies - Less costly to review prior studies rather creating new one - Time consuming and not necessarily easy to combine studies - Results are often inconclusive Meta-Analysis: - Statistically combines results from different studies (Could be part of a systematic review) - Subset of systematic reviews; combines qualitative and quantitative data from several studies to develop one conclusion that has greater statistical power - This information has confirmatory data analysis, can represent the population affected, and evidence-based - Con: time consuming to identify appropriate studies, req advanced stat techniques Increasing Evidence Strength: 1. Test tube research 2. Animal Research 3. Ideas, opinions, editorials 4. Case Reports/Series 5. Cross Sectional 6. Case Control 7. Cohort 8. RCT 9. Systematic Review 10. Meta Analyis TED TALK “BATTLING BAD SCIENCE” BY BEN GOLDACRE Art of Medical Decision Making: - Only a few cases in practice actually perfectly fit clinical studies in the literature o Many cases are unique - Clinicians should rely on Anecdotal Experience and local culture o Anecdotal Experience: ▪ Personal experience that can be measured only by that individual. It is subjective. o Local Culture: ▪ Culture of the place physician is practicing in; must be considerate of the cultural values and traditions o Both are tough to quantify the risks and benefits Article: Uncertainty is Hard for Doctors: - Patient is having abnormal heart rhythms and doctor is trying to decide whether or not to implant a deliberator. No decisive scientific evidence for him. Despite the uncertainty, the decision was made to implant the deliberator (Routine low-risk surgery). - Hell broke loose; Defibrillator shocked his heart at wrong times, patient and to take meds to lower heart rate and meds caused pooping, depression etc. - Defibrillator was turned off TED TALK- DOCTORS MAKE MISTAKES. CAN WE TALK ABOUT THAT? - BRIAN GOLDMAN Data Driven Decision Making: Clinical Decision Support System: - System that support decision making; application that analyzes data o help healthcare providers make clinical decisions - Adaptation of decision support system used to support business management - 2 major decision types in medicine: o Diagnosis o Treatment - This can be used for: o Knowledge retrieval (RODIA- Quantitative Eval. Of X-ray) , diagnostic assistance (DiagnosisPro,D xMate), treatment planning, and drug prescription (MYCIN- Antibiotic Rec.) IBM Watson: - Ultimate Natural Language Prcoessing Machine - Can Handle Big Data; Volume- Scale of Data, Velocity- Analysis of Streaming Data, Variety- Different forms of data, Veracity- Uncertainty of Data - Big Data- Large data sets that can be analyzed via computer to reveal patterns, trends etc. - Able to generate a reasonable hypothesis and assign quantitative levels of confidence to them; o Levels of Confidence: Expressed as a %, captures the accuracy (true mean) of the system being tested. Perform more and more tests on a system and you become increasingly become confident in predicting the result. TED TALK: WATSON, JEOPARDY and ME, the OBSOLETE KNOW-IT-ALL – KEN JENNINGS IBM Watson as a Clinical Decision Support System (CDSS) 1. Physician poses a clinical problem 2. Watson mines the patient’s med record, family history, journal articles, physician notes, the whole 9 yards 3. Hypotheses are generated and tested 4. Watson provides potential diagnoses/recommended treatments With Physician Watson, an expected shortage of oncologists can be rectified. Watson Struggles with: 1. Misinterpretation due to medical jargon, poor data quality 2. Deeper knowledge required Causation Vs. Correlation: Correlation: How strongly pairs of variables are related; does not automatically mean that the change in one variable is the cause of the change in the values of the other variable. Not direct cause. CANNOT BE CAUSATIVE; there is a trend but no evidence of causation Causation: One event is the result of the occurrence of the other event; can be predicted to cause and can cause due to evidence that we have. Limitations of Observational Data: - Can’t tell causation w/o randomization b/c there the randomization prevents confounding factors. Has to be a random study to predict causation - Known Unknowns: I do a study and I know this factor is study. IDK this factor, but it might be affecting my study - Unknown Unknown: There might be a factor that I don’t even know about Data In Action Article: - Many hospitals fail to diagnose a 6 year old - Mayo Clinic (369 Data Scientists) analyzed genomic data, medical records, and published literature - A rare gene mutation that 10 children in the world have Shared Decision Making: - An approach/system where clinicians and patients communicate w/ each other to make decisions; there is 2-way discussions and deliberations - Patients are encouraged to make informed decision based on using the best available evidence - Patients are at the CENTRE of medical decision making - SDM is seen as paradigm-shifting as Evidence-Based Medicine SDM is prominent because: - Patients want more info. and want to participate more in decision making - Patient Autonomy - SDM doesn’t imply patients MUST participate in decision making, but gives them to the chance to Patients Care about: - Living - Quality of Life - Able to do things they enjoy - Finances - Benefits/Risks of treatment Clinicians Care about: - Reducing Mortality - Hospital length of stay - Readmitting to hospital - Morbidity Challenges to SDM: 1. Lac of trust b//w patient and physician 2. Explaining scientific evidence to people is not easy 3. Don’t know how to personalize clinical guidelines 4. Imbalance of power b/w patients & physicians TED TALK “MEET E-PATIENT DAVE” – Dave deBronkart Quality Adjusted Life Years (QALY) - Looks at Length and Quality - Made initially for cost-effectiveness analysis ($$/QALY) QALY = (# of years of life) x (utility) - Perfect health: Utility = 1.0 - Being dead: utility = 0.0 - Living w/ conditions (wheelchair, blind, deaf etc.): Utility= 0.0 to 1.0 If utility is 0.5, living 1year w/ that condition is equal to living 0.5 perfectly healthy years How do you determine utilities? People assign subjective values to various health conditions 3 Methods: 1. Rating Scale 2. Standard Gambling 3. Time Trade-Off 1. Rating Scale: Subjective assessment of a health condition on a numerical scale a. 0 to 10, 0 being dead and 10 being prefect health, rate living as an amputee b. Subjective 2. Standard Gamble: Willingness to gamble to gain perfect health a. Choose b/w remaining ill and a therapy that leads to either perfect health or death 3. Time Trade-off: Willingness to trade off time for better health a. 20 years in perfect health or 30 years with a condition Problems with QALY: - Zero years is an absolute value but 0 utility is arbitrary (Unspecified value) o Comma or death? - Diff. people would assign diff QALY values to same condition - Value of living additional years and Utility is age-dependent TED TALK “HEALTH CARE SHOULD BE A TEAM SPORT” ERIC DISHMAN Population-level Decision: Limited healthcare resources: - Exacerbated/worsened by aging population - Resource allocation is more and more challenging Societal issues - Individual goals do not see eye to eye w/ population goals - Ethical issues arise Decision Support tools that integrate both EBM and SDM are needed Physicians should be provided with the best available evidence in order to answer patients’ questions Existing evidence should be presentable and easy to understand Accept the help of machines. Unit 5 Mobile Health: mHealth: Practice of medicine and public health supported by mobile devices, empowers patients and many applications in developing and developed countries. - Rise of mobile tech is due to advnaces in electronics and wireless communication o Smartphones, tables, wearable and portable sensors o Glucose monitors (connected to smartphone) ▪ Portable devices to test blood glucose concentration • Diabetics need to check atleast once a day • Some glucose monitoring devices have a accompanying app (My assignment one; one touch monitor) Allows to stor and visualize glucose measurements o Google contact lens- Measures glucose levels in tears ▪ LED light will show there is abnormal glucose levels ▪ Under development - o Blood Pressure Monitors- Vital for individuals w/ hypertension ▪ Inflatable cuff (Non-Invasive); Some come with accompanying smartphne app similar to glucose monitors o Holter Monitor- Ambulatory heart rhythm monitor ▪ Can record up to 2 weeks ▪ Lots of wires and bulky. Can be inconvenient o iRhythm- Used to determine arrhythmia diagnosis ▪ Records for up to 14 days ▪ No wires ▪ After recording, patients mail the device and a diagnostic report is generated. • Can detect early indications of stroke and therefore leading to reduced emergency visits o Vscan Pocket Ultrasound- Hand-held ultrasound device that provides great image quality (Can be compared to high-end systems) ▪ 1 hour of continuous scanning in one charge ▪ Cost effective alternative to comprehensive imaging tests TED TALK “THE WIRELESS FUTURE OF MEDICINE” ERIC TOPOL Article- Can mobile health technologies transform health care Mobile health is driven by 3 forces: 1. Unsustainable healthcare spending ( Saving costs to going to go ultrasounds, ECG etc.) 2. Rapid grown in wireless connectivity and technology 3. People have a need for INDIVIDUALIZED Medicine There are 30,000 to 90,000 health apps but only 100 have been reviewed by FDA Mobile Health allows for : - Self-diagnosis, monitoring, and tracking - Clinicians can track a bunch of patients MOST POPULAR MOBILE HEALTH APP- Fitness Apps - Health and Fitness Tracking: Steps, speed, heart rate, hydration perspiration, calorie count, nutritional value, sleep - Patient Monitoring: Blood Pressure, temperature, glucose level, posture, medicine, alcohol, sun exposure - Medical Examination: Respiratory rate, lung air volume, ECG, EEG, blood test, color test, urine test, ultrasound imaging Vinod Khosla: CEO of Sun Microsystems - Predicts many doctors will be replaced by machines - Due to mobile health o Researchers ask 40 cardiologist whether the patient should have surgery or not. Half said yes, half said no. So, a patient having surgery would be contingent on the cardiologist you have, which is bad. o 2 year late, 40% of those cardiologist chaged their mind Guest Lecture : Plinio Morita - UbiLab: o Breathe: Evaluating the impact of mobile health tech. on Asthma patients o Ecobee: Enabling IoT thermostat tech to Predict progression of mental health diseases and mobility-related diseases ▪ IoT- Internet of Things inter-networking of physical devices, vehicles with software, sensors, and network connectivity to enable these objects to collect and exchange data. o Health Ecosphere: Evaluating the pathways toward successful commercialization of mHealth, eHealth, and wearable technologies - Future of Healthcare: What we envision and what the reality is! - Wearable tech to phone to cloud to desktop to physician/technician/analyst - Zero-Effort Technoglogy: emerging class of technology hat requires little or no effort from the people who use it. It uses advanced techniques, such as computer vision, sensor fusion, decision-making/planning, and machine learning to operate through the collection, analysis, and application of data about the user in patient’s context. - beat: real-time wireless ECG streaming Tech Gap: 1. Zero-Effort monitoring of vital sign/sleep quality: Measures sleep quality, heart rate, and respiratory rate while sleeping 2. Existing Tech: Reliance on single sensor; low reliability 3. Sleep motion patterns: Characterizes motion in bed Our current scenario: 1. Lots of innovation in the market; consumer level technology is immense 2. Low data reliability; data created from these tech are less reliable than clinical grade tech 3. Isolated data silos (repository of fixed data that is under the control of one department and isolated from rest of the org.); Companies don’t communicate and not a lot of data integration 4. Immense untapped potential: Each data silo has limited potential to inform healthcare; if they were combine it would be limitless 5. We are only focused on the traditionally oriented healthcare tech; scales, heart rate cuffs, sleep and heart rate monitors Data Integration: Bringing clinical data and consumer level tech data to the same data lake - This would allow us to remotely monitor critical patients and test effectiv