Medical Education
Core IM Episode
https://www.coreimpodcast.com/2020/12/09/hoofbeats-struggles-with-clinical-reasoning/
Diagnostic reasoning: diagnose the problem Reasoning process: Generate hypotheses → gather data → create a problem representation → refine hypotheses → select a working diagnosis
Problems at each step will lead to error and can be improved with specific tasks:
- Generate hypotheses - noon conference style reports, humandx
- Gather data as guided by hypotheses - "Provide the learner a chief complaint, ask them for a tiered differential, then ask them to choose only five pieces of historical information and five physical exam findings to distinguish between the diagnoses they’re considering (Parsons et al, 2020).", "Provide the learner a printed history and physical and ask them to highlight only the information they believe is most relevant for building and narrowing their differential."
- Problem representation - "Deconstruct the problem representation into its three primary components– patient demographics and risk factors, temporal pattern expressed using semantic qualifiers, and key symptoms and findings."
- Refine hypotheses / Select working dx ???
###Ch 1
Rereading leads to the illusion of knowing the content (mastering the text, instead). People who don't quiz themselves overestimate how much they know. Importantly - testing is a more effective learning tool (active retrieval) not just a method of evaluation. Factual knowledge is a prerequisite for higher level synthesis.
###Ch 2
- challenging recall (ie testing) enhances retention more than passive rereading / reviewing (even though it is not always perceived as more durable)
- equally applicable to facts, complex concepts, problem-solving techniques, motor skills
- repetition increases durability, as well as increases speed of recall, applicability to wider settings
- practice testing = more realistic sense of knowledge, ability to identify misconceptions
(note: in medicine, 'use' aka rounds functions somewhat like practice tests.. has to feel 'low stakes')
###Ch 3
- The feeling of difficulty is likely what makes things stick, with spaced (long enough so that you've started to forget) reputation strengthening the networks
- Subjects should be interleaved to facilitate connection building.
- Variation - build a broad enough schema
- 'Familiarity Trap' = feeling that you know something so that it no longer needs to be retrieved -Block practice (e.g. running the same drill over and over) feels easier, which is why it doesn't work as well (despite seeming like it does). Cards must be in random order
- ==Reflection is a form of retrieval practice==
Note: interleaving contextual interferences (below) seems related to studying on subjects related to patients - more connections to tie to? There is also informal spaced repetition of subjects that keep coming up in clinical practice.. but there's no guarantee that it will maximize the timing with respect to forgetting.
###Ch 4
- Learning = encoding to short-term working memory, consolidating into a stabilized memory trace associated with meaning and tying to past experiences (Binge learning occurs in short term memory without consolidation), and retrieval updates the learning and enables you to apply it (e.g. strengthens retrieval routes and establishes retrieval cues)
- Often we don't forget things - we are just unable to recall them because the cues haven't been used - difficulty of retrieval practice correlates with entrenched recall (=desirable difficulty - if it's a little rusty, reconstructing the answer takes more effort.. despite feeling less confident).
- Desirability difficulties = spacing, interleaving, variation, generation, and some types of contextual interferences. Other difficulties may detract from learning.
- mental models = interrelated sets of ideas or sequences of motor skills.
- key point: varied practice conditions, interleaving encourage ability to discriminate and induction.. possibly a draw back of your own anki cards? (monotonous)
###Ch 5
- avoid illusion of knowing by:
- frequent low stakes quizing(calibrating your judgement)
- judging ease of explaining a concept to someone.
- (for others, give corrective feedback to help them calibrate)
###Ch 6
- High structure builders (ie find patterns in the learning and insert content into that structure) learn better - e.g. determining rules instead of memorizing examples. Comparing/contrasting different examples -Reflection may encourage structure building.
###Ch 7
- adopt growth mindset, avoid performance goals and embrace learning goals
- deliberate practice: directed to surpass current performance / knowledge ability (often not-enjoyable, requires a coach to identify weak points)
- use memory cues (e.g. memory palaces, mnemonics) to enhance recall of learned material 
Making Educational Content (e.g. videos, lectures, sims)
Cognitive load theory: transient nature of sensory memory -> filtered for temporary storage; processed in working memory (limited capacity) before either being forgotten or moved to long-term memory.
Strict 10 minute limit to passive components; chunk information into 3-4 minute segments.
Active teaching/learning: goal is for the learner to work through problems - much worse retention if passive. Examples: apply knowledge after questions or interactive scenarios, integrate pauses with questions
[] update teaching methods to involve frequent directed (specific) questions that generalize the concept to a related scenario