# Book Notes - Optimisation of Behavioural, Biobehavioral and Biomedical Intervention. The Multiphase Optimization Strategy. MOST. By Linda M Collins. ## Preface Why the need for this MOST framework? Because interventions are awesome, but the way of improving interventions is too slow. 1. The accumulation of a coherent base of scientific knowledge is slow. We don't know exactly what works. 2. There is a lack of emphasis on steady, incremental, measurable improvement of interventions unlike consumer products. 3. To have "evidence based" interventions are expensive and demanding and not scalable. ### What is MOST? - It is a framework focusing on optimising the intervention process, to make sure that it works first then we do RCT. - Multidisciplinary approach - combining ideas from Engineering, Statistics, Biostatistics, behavioural science. - It has three phases. - Preparation Phase - Optimisation Phase - Evaluation Phase And repeat. ![[Screenshot 2022-08-27 at 5.35.12 PM.png]] > Remind me of the #Design_Thinking process. ## Chapter 1 : Conceptual Introduction to the Multiphase Optimisation Strategy (MOST) - Define: What is an intervention? - A programme with the objective of improving and/or maintaining human health and well-being - There are different interventions - Behavioural interventions - Based on modification of affective, cognitive and behavoural function. - Biomedical interventions - Medicine, chemistry. - Biobehavioral - Combination of both Usually scientist want to do RCT, but it's expensive and slow. - Borrowing idea from Engineering, manufacturing. The goal is to achieve an optimised process/outcome within constraints. Repeating the process to continually improve it. - The case study in the book talk about optimising a manufacturing process to improve production. What components are needed, or not needed. Then an improve process is created. - Similar idea to #solution-focus. Unless we change something, then we expect the same result! > - What's the goal of HOPES? What's the point of treatment? - TCU rate? Compliance? How much patient knows, psychoeducation? ### What information needed for Optimisation? page 5. > Reminded me of #requirements collection and constrants in project management. - What are the size and direction of each component's effect? - Which components demonstrate a positive effect? - Going back to the First Principles. - A null effect? a negative effect? - Is the performance of one component affected by the presence or level of one or more components? ### Components There are the building blocks, ingredients. Lego blocks. 1. select components that works. That can contribute to the goal/effect 1. This will be based on clinical experience, knowledge, current science. 2. E.g "Psychoeducation" "Motivational Interviewing", "Peers Supports" can be components. 2. How each components affect other components 3. Different components associated with different cost 1. e.g Session with a therapist cost more then a text. 4. As soon as the process is completed, immediately work on improving the process. 1. Which intervention is immediately scalable? (page 7) 2. Don't include all the components at first. 3. Need to do experiment to gather information - That's the Experimental designs not RCT. (Chapter 6) > How to know which component works? Is it like A/B Testing? >So far i am thinking, so, what's the intervention components for treatment for psychosis? > 1. Psychoeducation? > 2. Medication compliance? > 3. ... >What are the current evidence for treatment? ### Definition of Optimisation of an intervention. What's optimisation? "Optimization of an intervention is the process of identifying an intervention that provides the best expected outcome obtainable within key constraints imposed by the need for efficiency, economy, and/or scalability." > This is Kaizen. Keep improving the previous model. Bring the cost down. Increase effectiveness. - Because now we know which component works best, we can build on what we know. - The body of knowledge start to accumulate. - Four Desiderata: Effectiveness, Efficiency, Economy, and Scalability. (Why we want to keep improving) - But sometimes effectiveness will draft the rest. Think of the COVID-19 ART test kit or Vaccine. It could be very expensive initially, but cheaper as processes improves. ### 1.4 Resource Management Principle ![[Screenshot 2022-08-27 at 5.31.56 PM.png]] ### 1.6.2 Components "an intervention component is any part of an intervention that can be separated out for study" - candidate component is a component from which a level (e.g., no or yes, off or on, low intensity or high intensity) is to be selected for inclusion in an intervention. This term may refer to any type of component. - Content Component - Is the "what". For example, in Dr. B’s hypothetical ART adherence intervention described above, motivational interviewing, peer mentoring, text messaging, mind- fulness meditation, and behavioral skills training are all content components - Engagement / adherence Components - Stuff that keep participants engaged on the interventions. > We don't have that in HOPES now. - Fidelity components are aimed at maintaining a high level of fidelity of interven- tion delivery. Such components are usually aimed at those who deliver the interven- tion or the environment in which the intervention is to be delivered, rather than the individuals who are the target of the intervention. - How to prepare the environment to roll out the interventions (school, hospital etc) - Components for the environment, to encourage adoption. ## Chapter 2: Preparation Phase ### Conceptual Model - Engine that drive interventions - According to the W.K. Kellogg Foundation (2004), “[a] program logic model links outcomes (both short- and long-term) with program activities/processes and the theoretical assumptions/principles of the program” (p. iii). - The more specific a conceptual model, the better. It offer high level of specificity in three areas: - 1. Describe the casual process to be intervened. - 2. Express the investigator's approach to intervening on these casual process. The strategy - 3. Detail any anticipated moderation of the effects one component have on another (How Component A may affect B) Uses of Conceptual Model - Guide decision making - Highlight redundancy - Are all the components coherent? - Also can see the moderate effects. What other factors can influence it, like SES, education, Gender. ![[Screenshot 2022-08-27 at 5.40.39 PM.png]] ### Tracing casual chain - Causal chain: Used in behavioral biobehavioral sciences. - ==If the primary outcome is far into the future. We can evaluate the components, moderators. Since if we can change the moderator, then it should change the primary outcome.== ^9ed9d3 Conceptual Model is a living document. - Try to have one components to one variables for better, cleaner control. If not we don't know which works. - What might moderate the effect on components. For example, depression can moderate the effects of mindfulness. ![[Screenshot 2022-08-27 at 5.44.19 PM.png]] ### 2.10 Optimisation Criterion What guide the optimisation process is 1. Does it improve outcome? 2. Is it within constraints - cost? 3. It should be operationalised, in numbers.. .. so that can be measured. Set the criterion early during the preparation phase, so that we can plan how to capture the data needed for optimisation. ## Chapter 3 - 7 Explained the statistics methods. Factorial, trials that can be used, that are not RCT. ANOVA... i don't understand. ## Chapter 8 Adaptive Intervention - Relavant to HOPES - Because heterogeneity in responses, and participants. One size doesn't fit all - Patients that responded well don't need as much resources as those who are not responding. - Idea of having decision points, and to have different interventions based on responses. - ![[Screenshot 2022-08-27 at 5.50.13 PM.png]] But still got lots of stats.. don't understand.