PK C3]_rels/PK C3] docProps/PK C3]ppt/PK C3] ppt/_rels/PK C3] ppt/charts/PK C3]ppt/charts/_rels/PK C3]ppt/embeddings/PK C3] ppt/media/PK C3]ppt/slideLayouts/PK C3]ppt/slideLayouts/_rels/PK C3]ppt/slideMasters/PK C3]ppt/slideMasters/_rels/PK C3] ppt/slides/PK C3]ppt/slides/_rels/PK C3] ppt/theme/PK C3]ppt/notesMasters/PK C3]ppt/notesMasters/_rels/PK C3]ppt/notesSlides/PK C3]ppt/notesSlides/_rels/PK C3]LU%)%)[Content_Types].xml PK C3]]] _rels/.rels PK C3]nؔttdocProps/app.xml 0 0 Microsoft Office PowerPoint On-screen Show (16:9) 0 20 20 0 0 false Fonts Used 2 Theme 1 Slide Titles 20 Arial Calibri Office Theme Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20 PptxGenJS false false false 16.0000 PK C3]ˌdocProps/core.xml PptxGenJS Presentation PptxGenJS Presentation PptxGenJS PptxGenJS 1 2026-08-23T06:26:06Z 2026-08-23T06:26:06Z PK C3]bbppt/_rels/presentation.xml.rels PK C3]Oݨ ppt/theme/theme1.xmlPK C3]֊:ppt/presentation.xml PK C3]Xppt/presProps.xml PK C3]ppt/tableStyles.xml PK C3]D >00ppt/viewProps.xml PK C3]H7t!ppt/slideLayouts/slideLayout1.xml PK C3]ђ77,ppt/slideLayouts/_rels/slideLayout1.xml.rels PK C3]ppt/slides/slide1.xml AI IN DAILY MEDICAL PRACTICEModule 4Critical Appraisal, Bias & Clinical AutonomySelf-paced + seminar, workshop & simulationAI in Daily Medical Practice — a curriculum for practicing physiciansPK C3]3 ppt/slides/_rels/slide1.xml.rels PK C3].ppt/notesSlides/notesSlide1.xml 1PK C3]:A*ppt/notesSlides/_rels/notesSlide1.xml.rels PK C3]|@G@Gppt/slides/slide2.xml BEFORE YOU STARTHow this module works1Preview Quiz5 min, ungraded. Orients you before reading.→2Handout / SlidesSection B needs your Module 2 and 3 outputs handy. Section E is a worked simulation — answer it yourself first.→3Seminar, Workshop & SimulationLive sessions with your cohort — compare notes, practice the routine, and work the high-stakes case together.→4Post-Test12 questions, scored. 80% pass, retake anytime.2PK C3]2- ppt/slides/_rels/slide2.xml.rels PK C3]ppt/notesSlides/notesSlide2.xml 2PK C3]xշ*ppt/notesSlides/_rels/notesSlide2.xml.rels PK C3]f8)8)ppt/slides/slide3.xml BY THE END OF THIS MODULEYou will be able to1Identify hallucination, overconfidence, and demographic or algorithmic bias in real AI outputs.2Apply a structured routine for interrogating AI outputs — especially when they confirm your own initial impression.3Articulate the legal and professional boundary of physician accountability when using AI-assisted tools.3PK C3]W/ ppt/slides/_rels/slide3.xml.rels PK C3]K |Őppt/notesSlides/notesSlide3.xml 3PK C3]9 Y*ppt/notesSlides/_rels/notesSlide3.xml.rels PK C3]Bjjppt/slides/slide4.xml ASECTION AWhy This Module Exists:The Confirmation-Bias TrapKNOW4PK C3]` ppt/slides/_rels/slide4.xml.rels PK C3]vsppt/notesSlides/notesSlide4.xml 4PK C3]J *ppt/notesSlides/_rels/notesSlide4.xml.rels PK C3]V,-,-ppt/slides/slide5.xml SECTION A · TWO NAMED PATTERNSThe trap isn't carelessness — it's feeling in control5ThesovereigntytrapThe comfortable but false sense that you remain fully in control simply because you're the one reviewing the AI's output — reviewing isn't the same as scrutinizing.ThehollowedmindWhat happens when review becomes routine and passive — the skill of independent judgment erodes precisely because it's exercised less, not because it's needed less.THE WORKING RULE FOR THIS MODULEOversight that isn't effortful isn't real oversight. A checkbox ticked without genuine scrutiny protects no one — not the patient, and not you.PK C3]5 ppt/slides/_rels/slide5.xml.rels PK C3]W8ppt/notesSlides/notesSlide5.xml 5PK C3]Qe*ppt/notesSlides/_rels/notesSlide5.xml.rels PK C3]#wvvppt/slides/slide6.xml BSECTION BSpotting Hallucination andOverconfidence in Your Own OutputsTHINK6PK C3]ج+ ppt/slides/_rels/slide6.xml.rels PK C3]zppt/notesSlides/notesSlide6.xml 6PK C3]=|*ppt/notesSlides/_rels/notesSlide6.xml.rels PK C3]s88ppt/slides/slide7.xml SECTION B · SELF-AUDIT WORKSHEETTurn the lens on outputs you already trusted7Pull up the AI-drafted note from Module 2 Section C (the hypertension follow-up) and the AI-generated differential from Module 3 Section D (the Dili fever case).1Is there any detail stated with confidence that you can't actually verify against the source encounter?2Did anything in the output sound plausible only because it was phrased plausibly — not because you checked it?3Where did the output most closely match what you already expected — and did that agreement make you look closer, or less close?4If you re-did this today with a more critical eye, what would you flag that you didn't flag the first time?PK C3]F ppt/slides/_rels/slide7.xml.rels PK C3])lppt/notesSlides/notesSlide7.xml 7PK C3]|g*ppt/notesSlides/_rels/notesSlide7.xml.rels PK C3] Mmyyppt/slides/slide8.xml CSECTION CAlgorithmic and Demographic Bias:How It Gets Into Clinical ToolsKNOW8PK C3]6 ppt/slides/_rels/slide8.xml.rels PK C3]iސppt/notesSlides/notesSlide8.xml 8PK C3]pO*ppt/notesSlides/_rels/notesSlide8.xml.rels PK C3]V N+N+ppt/slides/slide9.xml SECTION C · THREE SOURCESBias isn't a bug added later — it's built in from the start9Training dataimbalanceThe tool performs best on the populations best represented in its training data — and worse on everyone else, including, likely, your patients.Historical biasbaked into labelsIf the historical data used to train the tool reflects past inequities in who got tested, diagnosed, or treated, the tool learns and repeats that pattern.ProxyvariablesA variable that correlates with — but isn't the same as — what the tool is actually trying to predict, quietly encoding bias into an otherwise neutral-looking output.PK C3]>$ ppt/slides/_rels/slide9.xml.rels PK C3]qppt/notesSlides/notesSlide9.xml 9PK C3]1*ppt/notesSlides/_rels/notesSlide9.xml.rels PK C3]-uppt/slides/slide10.xml SECTION C · WHY THIS MATTERS HERETimor-Leste is rarely the population a tool was validated on10Most widely available clinical AI tools were developed and validated on populations, health systems, and disease patterns unlike Timor-Leste's. That gap doesn't make a tool useless — but it means its outputs need more scrutiny here, not less, and the burden of catching the mismatch falls on you.PK C3]Ѳ!ppt/slides/_rels/slide10.xml.rels PK C3]O ppt/notesSlides/notesSlide10.xml 10PK C3]T+ppt/notesSlides/_rels/notesSlide10.xml.rels PK C3]iippt/slides/slide11.xml DSECTION DWorkshop: The “Open theBlack Box” RoutineTHINK11PK C3];!ppt/slides/_rels/slide11.xml.rels PK C3]s6ӑ ppt/notesSlides/notesSlide11.xml 11PK C3]O+ppt/notesSlides/_rels/notesSlide11.xml.rels PK C3]3F3Fppt/slides/slide12.xml SECTION D · ONE ROUTINE, NAMED ONCEEvery habit this course has built is this routine in different clothes121State your own answer first — before you read the AI's output, what would you have said?2Ask what it's actually based on — do you know, or are you assuming it's sound because it sounds sound?3If it agrees with you, look harder, not less — agreement is exactly when scrutiny naturally drops.4Treat unexplained precision as a flag, not reassurance — verify it, don't trust it because it's specific.5Name one fact the tool couldn't have known — local context, exposure history, exam findings.6Decide, document, and own the call — the record should show your reasoning, not “AI recommended X.”PK C3]c!ppt/slides/_rels/slide12.xml.rels PK C3]*)@ ppt/notesSlides/notesSlide12.xml 12PK C3]Fb+ppt/notesSlides/_rels/notesSlide12.xml.rels PK C3]{{ppt/slides/slide13.xml ESECTION ESimulation: A High-Stakes CaseWhere the AI Is Confidently WrongTHINK13PK C3]x!ppt/slides/_rels/slide13.xml.rels PK C3]Ї ppt/notesSlides/notesSlide13.xml 13PK C3]yv+ppt/notesSlides/_rels/notesSlide13.xml.rels PK C3])>ppt/slides/slide14.xml SECTION E · STEP 1 — READ THIS FIRSTForm your own answer before reading on14CASEA 6-year-old, 18 kg, with known stage 3 chronic kidney disease presents febrile with a UTI at the end of a long shift. Renal dosing for the aminoglycoside you're considering isn't something you have memorized cold, so you ask a general-purpose AI chatbot to confirm the dose.The chatbot confidently states the standard paediatric dose, and adds, unprompted and without a citation, that “a 10% dose reduction is generally sufficient” for renal impairment at this stage.Before turning the page — what's your own read on that answer?PK C3]O!ppt/slides/_rels/slide14.xml.rels PK C3]0 ppt/notesSlides/notesSlide14.xml 14PK C3]?ݤ+ppt/notesSlides/_rels/notesSlide14.xml.rels PK C3]Mʧ33ppt/slides/slide15.xml SECTION E · DISCUSSIONWhat this case is built to show151The specific, uncited “10%” is the hallucination “tell” — not a reassuring level of detail. Confident precision without a source is a flag.2Renal dosing adjustment in children isn't a simple linear percentage calculation — it depends on the drug, the degree of impairment, and the specific regimen.3The sovereignty trap is strongest exactly when you're tired at the end of a shift and want the decision to be over.4The correct move is always to verify against a real paediatric renal-dosing reference or a pharmacist before acting — this case deliberately does not hand you a “correct” dose to memorize.PK C3]*R!ppt/slides/_rels/slide15.xml.rels PK C3]Qz ppt/notesSlides/notesSlide15.xml 15PK C3]$Q+ppt/notesSlides/_rels/notesSlide15.xml.rels PK C3] ]}}ppt/slides/slide16.xml FSECTION FLegal and Professional Accountability:Where the Line Actually SitsKNOW16PK C3] !ppt/slides/_rels/slide16.xml.rels PK C3]  ppt/notesSlides/notesSlide16.xml 16PK C3] y+ppt/notesSlides/_rels/notesSlide16.xml.rels PK C3]eg..ppt/slides/slide17.xml SECTION F · THE PRINCIPLEUsing an AI tool doesn't shift accountability off the physician17The decision remains yours. The tool is an input you consulted — not a co-signer who shares the liability.1The standard of care you're held to is unchanged by whether an AI tool was involved.2Documentation should reflect your reasoning — not delegate the decision to the tool on paper.3Most jurisdictions, including Timor-Leste, lack AI-specific liability frameworks — ordinary professional standards apply in full.PK C3]檋!ppt/slides/_rels/slide17.xml.rels PK C3]r. ppt/notesSlides/notesSlide17.xml 17PK C3]BI+ppt/notesSlides/_rels/notesSlide17.xml.rels PK C3]U[[ppt/slides/slide18.xml SECTION F · WHAT'S NEXTThe regulatory detail comes next18This module deliberately stays at the level of general professional principle. Timor-Leste-specific regulatory detail — what AIFAESA and other local authorities expect, and how it applies to your practice — is the subject of Module 5.PK C3]Yh!ppt/slides/_rels/slide18.xml.rels PK C3] ppt/notesSlides/notesSlide18.xml 18PK C3]N+ppt/notesSlides/_rels/notesSlide18.xml.rels PK C3]" ppt/slides/slide19.xml BEFORE YOU MOVE ONModule 4 summaryKNOWThe sovereignty trap and the hollowed mind; how training-data imbalance, historical bias, and proxy variables put bias into clinical tools; where physician accountability actually sits and doesn't shift.THINKThe Black Box Routine as one habit underneath everything this course has built, practiced on your own past outputs and on a high-stakes simulation where the AI is confidently wrong.19PK C3]~!ppt/slides/_rels/slide19.xml.rels PK C3]0[ ppt/notesSlides/notesSlide19.xml 19PK C3]i+ppt/notesSlides/_rels/notesSlide19.xml.rels PK C3]8m>ppt/slides/slide20.xml NEXT STEPSTake the Post-Test12 scored questions, 80% pass mark. After the seminar, workshop, and simulation sessions, and once Module 4 is evaluated, Module 5 — Ethics, Governance & the Local Regulatory Context — opens next.Module 4 of 6 — Critical Appraisal, Bias & Clinical Autonomy20PK C3]g_!ppt/slides/_rels/slide20.xml.rels PK C3]7 ppt/notesSlides/notesSlide20.xml 20PK C3]SB+ppt/notesSlides/_rels/notesSlide20.xml.rels PK C3]K !ppt/slideMasters/slideMaster1.xml PK C3]N),ppt/slideMasters/_rels/slideMaster1.xml.rels PK C3]6TT!ppt/notesMasters/notesMaster1.xml 7/23/19Click to edit Master text stylesSecond levelThird levelFourth levelFifth level‹#›PK C3]s **,ppt/notesMasters/_rels/notesMaster1.xml.rels PK C3]_rels/PK C3] $docProps/PK C3]Kppt/PK C3] mppt/_rels/PK C3] ppt/charts/PK C3]ppt/charts/_rels/PK C3]ppt/embeddings/PK C3] ppt/media/PK C3]Bppt/slideLayouts/PK C3]qppt/slideLayouts/_rels/PK C3]ppt/slideMasters/PK C3]ppt/slideMasters/_rels/PK C3]  ppt/slides/PK C3]3ppt/slides/_rels/PK C3] bppt/theme/PK C3]ppt/notesMasters/PK C3]ppt/notesMasters/_rels/PK C3]ppt/notesSlides/PK C3]ppt/notesSlides/_rels/PK C3]LU%)%)P[Content_Types].xmlPK C3]]] ,_rels/.relsPK C3]nؔtt,/docProps/app.xmlPK C3]ˌ6docProps/core.xmlPK C3]bb9ppt/_rels/presentation.xml.relsPK C3]Oݨ Hppt/theme/theme1.xmlPK C3]֊:ippt/presentation.xmlPK C3]Xxppt/presProps.xmlPK C3]&zppt/tableStyles.xmlPK C3]D >00 {ppt/viewProps.xmlPK C3]H7t!l~ppt/slideLayouts/slideLayout1.xmlPK C3]ђ77,Zppt/slideLayouts/_rels/slideLayout1.xml.relsPK C3]ۂppt/slides/slide1.xmlPK C3]3 ppt/slides/_rels/slide1.xml.relsPK C3].ppt/notesSlides/notesSlide1.xmlPK C3]:A*ˣppt/notesSlides/_rels/notesSlide1.xml.relsPK C3]|@G@Gޥppt/slides/slide2.xmlPK C3]2- Qppt/slides/_rels/slide2.xml.relsPK C3]]ppt/notesSlides/notesSlide2.xmlPK C3]xշ**ppt/notesSlides/_rels/notesSlide2.xml.relsPK C3]f8)8)=ppt/slides/slide3.xmlPK C3]W/ !ppt/slides/_rels/slide3.xml.relsPK C3]K |Ő#ppt/notesSlides/notesSlide3.xmlPK C3]9 Y**ppt/notesSlides/_rels/notesSlide3.xml.relsPK C3]Bjj,ppt/slides/slide4.xmlPK C3]` 1Dppt/slides/_rels/slide4.xml.relsPK C3]vs=Fppt/notesSlides/notesSlide4.xmlPK C3]J * Mppt/notesSlides/_rels/notesSlide4.xml.relsPK C3]V,-,-Oppt/slides/slide5.xmlPK C3]5 ||ppt/slides/_rels/slide5.xml.relsPK C3]W8~ppt/notesSlides/notesSlide5.xmlPK C3]Qe*Uppt/notesSlides/_rels/notesSlide5.xml.relsPK C3]#wvvhppt/slides/slide6.xmlPK C3]ج+ ppt/slides/_rels/slide6.xml.relsPK C3]zppt/notesSlides/notesSlide6.xmlPK C3]=|*ppt/notesSlides/_rels/notesSlide6.xml.relsPK C3]s88ppt/slides/slide7.xmlPK C3]F ppt/slides/_rels/slide7.xml.relsPK C3])lppt/notesSlides/notesSlide7.xmlPK C3]|g*ppt/notesSlides/_rels/notesSlide7.xml.relsPK C3] Mmyyppt/slides/slide8.xmlPK C3]6 xppt/slides/_rels/slide8.xml.relsPK C3]iސppt/notesSlides/notesSlide8.xmlPK C3]pO*Qppt/notesSlides/_rels/notesSlide8.xml.relsPK C3]V N+N+dppt/slides/slide9.xmlPK C3]>$ ;ppt/slides/_rels/slide9.xml.relsPK C3]q=ppt/notesSlides/notesSlide9.xmlPK C3]1*Dppt/notesSlides/_rels/notesSlide9.xml.relsPK C3]-uFppt/slides/slide10.xmlPK C3]Ѳ!Yppt/slides/_rels/slide10.xml.relsPK C3]O [ppt/notesSlides/notesSlide10.xmlPK C3]T+bppt/notesSlides/_rels/notesSlide10.xml.relsPK C3]iidppt/slides/slide11.xmlPK C3];!P|ppt/slides/_rels/slide11.xml.relsPK C3]s6ӑ ^~ppt/notesSlides/notesSlide11.xmlPK C3]O+-ppt/notesSlides/_rels/notesSlide11.xml.relsPK C3]3F3FBppt/slides/slide12.xmlPK C3]c!ppt/slides/_rels/slide12.xml.relsPK C3]*)@ ppt/notesSlides/notesSlide12.xmlPK C3]Fb+ppt/notesSlides/_rels/notesSlide12.xml.relsPK C3]{{ppt/slides/slide13.xmlPK C3]x!Jppt/slides/_rels/slide13.xml.relsPK C3]Ї Xppt/notesSlides/notesSlide13.xmlPK C3]yv+'ppt/notesSlides/_rels/notesSlide13.xml.relsPK C3])><ppt/slides/slide14.xmlPK C3]O!ppt/slides/_rels/slide14.xml.relsPK C3]0 ppt/notesSlides/notesSlide14.xmlPK C3]?ݤ+ ppt/notesSlides/_rels/notesSlide14.xml.relsPK C3]Mʧ33"ppt/slides/slide15.xmlPK C3]*R!Vppt/slides/_rels/slide15.xml.relsPK C3]Qz Xppt/notesSlides/notesSlide15.xmlPK C3]$Q+_ppt/notesSlides/_rels/notesSlide15.xml.relsPK C3] ]}}appt/slides/slide16.xmlPK C3] !lyppt/slides/_rels/slide16.xml.relsPK C3]  z{ppt/notesSlides/notesSlide16.xmlPK C3] y+Ippt/notesSlides/_rels/notesSlide16.xml.relsPK C3]eg..^ppt/slides/slide17.xmlPK C3]檋!ppt/slides/_rels/slide17.xml.relsPK C3]r. !ppt/notesSlides/notesSlide17.xmlPK C3]BI+ppt/notesSlides/_rels/notesSlide17.xml.relsPK C3]U[[ppt/slides/slide18.xmlPK C3]Yh!ppt/slides/_rels/slide18.xml.relsPK C3] ppt/notesSlides/notesSlide18.xmlPK C3]N+qppt/notesSlides/_rels/notesSlide18.xml.relsPK C3]" ppt/slides/slide19.xmlPK C3]~!ppt/slides/_rels/slide19.xml.relsPK C3]0[ ppt/notesSlides/notesSlide19.xmlPK C3]i+ppt/notesSlides/_rels/notesSlide19.xml.relsPK C3]8m>ppt/slides/slide20.xmlPK C3]g_!ppt/slides/_rels/slide20.xml.relsPK C3]7 ppt/notesSlides/notesSlide20.xmlPK C3]SB+'ppt/notesSlides/_rels/notesSlide20.xml.relsPK C3]K !)ppt/slideMasters/slideMaster1.xmlPK C3]N),Hppt/slideMasters/_rels/slideMaster1.xml.relsPK C3]6TT!Jppt/notesMasters/notesMaster1.xmlPK C3]s **,:hppt/notesMasters/_rels/notesMaster1.xml.relsPKssf!i