高醫(yī)大腎臟內(nèi)科

HowtoMakeHypotheses顧進(jìn)裕高醫(yī)大腎臟內(nèi)科,啟川大樓14ES07-3121101EXT.7353~12HowtoMakeHypothesesIntroductionResearchLogicCausationHypothesisDiscussion不知道與好奇心生下了為什麼為什麼與數(shù)學(xué)生下了科學(xué)科學(xué)與藝術(shù)生下了醫(yī)學(xué)RenéDescartes

(March31,1596–February11,1650)我思故我在

(Cogito,ergosum)懷疑是智慧的源頭:讀書(shū)時(shí)，不要急著辯駁，也不要盲目信從，或僅僅想找一個(gè)茶餘飯後的話(huà)題，而是要去思考和判斷。有些書(shū)淺嚐即可，有些書(shū)不妨狼吞虎嚥，有些書(shū)卻需要細(xì)細(xì)嚼、慢慢嚥，換句話(huà)說(shuō)，有些書(shū)只需要選讀，有些書(shū)只需要瀏覽，有些書(shū)卻需要從頭到尾，仔細(xì)地體悟與省思。──FrancisBaconThedifficultylies,notinthenewideas,butinescapingtheoldones.-JohnMaynardKeynes,Englisheconomist創(chuàng)造力的來(lái)源遺傳努力工作(geniusis99%perspiration)知識(shí)(FrancisBacon)忘記學(xué)校所教的(H.P.);遊戲蘋(píng)果;在沙灘撿石頭;站在巨人肩上洗澡犯錯(cuò)(Penicillin)看星星(PCR)做夢(mèng)(aromatic)修行(朱銘),頓悟QuantumleapParadigmshiftIfatfirsttheideaisnotabsurd,thenthereisnohopeforit.Imaginationismoreimportantthanknowledge.Educationiswhatremainsafteronehasforgotteneverythinghelearnedinschool.Theimportantthingisnottostopquestioning.Curiosityhasitsownreasonforexisting.Anyonewhohasnevermadeamistakehasnevertriedanythingnew.SirArthurConanDoyle,1859-1930InspectorGregory:“Isthereanyotherpointtowhichyouwouldwishtodrawmyattention?〞Holmes:“Tothecuriousincidentofthedoginthenight-time.〞“Thedogdidnothinginthenighttime.〞“Thatwasthecuriousincident,〞r(shí)emarkedSherlockHolmes.-TheadventureofSilverBlazeGoogleGooglescholarPubmedWikimediaLibraryofKMUHeuristics(啟發(fā)式思考)1.Ifyouarehavingdifficultyunderstandingaproblem,trydrawingapicture.2.Ifyoucan'tfindasolution,tryassumingthatyouhaveasolutionandseeingwhatyoucanderivefromthat("workingbackward").3.Iftheproblemisabstract,tryexaminingaconcreteexample.4.Trysolvingamoregeneralproblemfirst(the"inventor'sparadox":themoreambitiousplanmayhavemorechancesofsuccess).-HowtoSolve

科學(xué)理論的標(biāo)準(zhǔn)清晰及合符邏輯(LogicalandClear)內(nèi)部一致(InternalConsistency)提供可驗(yàn)證的假說(shuō)(TestableHypothesesCanbeDeduced)可被推翻和否認(rèn)的(CanbeDisconfirmedorFalsified)通則性(Generalizability)簡(jiǎn)約的(Parsimonious):Occam’srazor理論及假說(shuō)的建構(gòu)小結(jié)真正的科學(xué)研究必由理論出發(fā)，而以驗(yàn)證及改進(jìn)理論為終如果尚未建構(gòu)清楚的理論模型及假說(shuō)，不應(yīng)以「碰運(yùn)氣」的方式從實(shí)證的資料中看那些變項(xiàng)呈相關(guān)，然後撰寫(xiě)報(bào)告，我們稱(chēng)這樣的做法為「在資料中採(cǎi)礦」(datamining)，是不應(yīng)該的，因?yàn)檫@與科學(xué)研究的本意相違背如果我們?cè)隍?yàn)證理論時(shí)，從資料中發(fā)現(xiàn)意外的現(xiàn)象，那就作為日後進(jìn)一步驗(yàn)證新的假說(shuō)的基礎(chǔ)HowtoMakeHypothesesIntroductionResearchLogicCausationHypothesisDiscussion研究的種類(lèi)A:Hypothesis-drivenDiscovery-drivenMethodB:基礎(chǔ)研究:分子、細(xì)胞、動(dòng)物臨床研究:診斷或篩檢、治療、預(yù)後、病因或副作用C:相關(guān):Observational因果:Experimental研究問(wèn)題

(ResearchProblem)重要性:那一個(gè)最重要?貢獻(xiàn)性原創(chuàng)性常見(jiàn)性花錢(qián)性致病性與致命性相關(guān)性影響性知識(shí)性與理論性實(shí)用性可行性Goals&ObjectivesGoal:Long-termgoaltosolvetheproblemObjectives(specificaims):Short-termobjectivestosolvethespecificproblemsoranswerthequestions(hypotheses).Itshouldbelogical(inevitablefromthebackground)Exploration,description,hypothesistesting,casestudy‘Wouldyoutellme,please,whichwayIoughttogofromhere?'saidAlice.`Thatdependsagooddealonwhereyouwanttogetto,'saidtheCat.

`Idon'tmuchcarewhere--'saidAlice.

`Thenitdoesn'tmatterwhichwayyougo,'saidtheCat.`--solongasIgetSOMEWHERE,'Aliceaddedasanexplanation.

`Oh,you'resuretodothat,'saidtheCat,`ifyouonlywalklongenough.'Goals&Objectives醫(yī)學(xué)論文各節(jié)的功能題目與摘要:敘述問(wèn)題與答案前言:為什麼(Why)問(wèn)這個(gè)問(wèn)題材料與方法:如何(How)解決這個(gè)問(wèn)題結(jié)果與圖表:發(fā)現(xiàn)什麼(What)去解決這個(gè)問(wèn)題討論:提出並解釋對(duì)於這個(gè)問(wèn)題的答案參考文獻(xiàn):

我站在巨人的肩上,所以(對(duì)於這個(gè)問(wèn)題)能看得更遠(yuǎn)-牛頓Hypotheses

orQuestions(1)Themostimportantstatementin“Introduction〞Asuper-topicsentence(inpresenttense)Sometimesstatedasobjectives(specificaims)Logical(inevitable)fromthebackground(everyquestionmusthaveabackground)Hypotheses

orQuestions(2)Ifquestionhasbeenstatedcompletelyintheunknownin“Background〞:Toanswerthisquestion,…Ifquestionhasbeenstatedonlypartiallyintheunknown:Therefore(Hence),wetestedthehypothesisthat…weasked(thequestion)whether…we(attemptedto)determine…werestudiedtoaddressXquestions:First,…Ifquestionistobestatedasanobjective:Todeterminewhether(which)…,we…HowtoMakeHypothesesIntroductionResearchLogicCausationHypothesisDiscussion探究方法定義概念範(fàn)例歸納法組合各明確且特定的單項(xiàng)概念，求出其關(guān)聯(lián)性而歸結(jié)出一個(gè)總括性的概念現(xiàn)象觀察→綜合歸納→導(dǎo)出理論觀察人口流動(dòng)現(xiàn)象：(1)經(jīng)濟(jì)繁榮的城市人口移出率不高。(2)經(jīng)濟(jì)衰退的鄉(xiāng)村人口移出率很高?！C合歸納出：「貧窮」可能是造成人口移出的主要因素。→導(dǎo)出人口推拉理論：鄉(xiāng)村多貧窮，形成人口移入的「推力」；城市因就業(yè)機(jī)會(huì)多，經(jīng)濟(jì)發(fā)達(dá)，形成人口移入的「拉力」。演繹法由較總括性的概念推演出一個(gè)明確、特定且有關(guān)聯(lián)的單項(xiàng)概念引用理論→演繹推論→現(xiàn)象驗(yàn)證引用地震發(fā)生理論：(1)地震是地殼受外力擠壓而產(chǎn)生的震動(dòng)。(2)板塊碰撞帶因地殼受碰撞而擠壓，容易發(fā)生地震?！堇[邏輯：臺(tái)灣位於板塊碰撞帶，地震頻繁，應(yīng)可發(fā)現(xiàn)火山和斷層等地震所產(chǎn)生的地形特徵?！鷮?shí)察驗(yàn)證：進(jìn)行野外實(shí)察以驗(yàn)證假說(shuō)。表1歸納法與演繹法比較LogicalreasoningGivenpreconditionsα,postconditionsβandtheruleR1:α∴β(αthereforeβ).Deduction:determiningβ.Itisusingtheruleanditspreconditionstomakeaconclusion(α∧R1?β).Induction:determiningR1.ItislearningR1afternumerousexamplesofβandα.Abduction(溯因):Inferencetothebestexplanation.Onechooseshypothesiswhichwould,iftrue,bestexplaintheevidence(β∧R1?α).InductiveReasoningGeneralization:Mycatlikestuna,hiscatlikestuna,etc.AllormostcatsliketunaStatisticalsyllogism:Mostcatsliketuna.SlothisacatSlothlikestunaSimpleinduction:AllofthecatsI’veknownliketunaThisnewcatwillliketunaAnalogy:JimandBillhavesimilarcharacteristics.JimislovedbyhisclassmatesBillislikedbyhisclassmatesDeductiveReasoning(I)DeductiveReasoning(II)所有的藥物都有副作用,假設(shè)某甲宣稱(chēng)他的新藥沒(méi)有副作用此話(huà)為真,則邏輯上如何?某奶粉公司添加抗體,宣稱(chēng)能預(yù)防病菌?某酵素能在試管中化痰,請(qǐng)問(wèn)該酵素口服後是否可能有效?某科學(xué)家訓(xùn)練一隻跳蚤在聽(tīng)到大聲後便會(huì)跳高,當(dāng)他把該跳蚤的雙腿剪斷後,跳蚤在聽(tīng)到大聲後卻不會(huì)跳高,因此證明了跳蚤是用雙腿聽(tīng)聲音?Logic(1)科學(xué)論文是由論證(argument)所構(gòu)成的論證=前提(premise)+推論(inference)+結(jié)論(conclusion)有效的(valid)論證:前提蘊(yùn)含結(jié)論,假設(shè)你接受前提,則你必須接受結(jié)論真確的(sound)論證:“前提是真〞的有效論證Logic(2):假設(shè)p53基因突變(P)則有癌癥(Q)演繹法(deduction,數(shù)學(xué)上的必然):一定對(duì)某甲的p53基因有突變,因此有癌癥某乙沒(méi)有癌癥,因此他的p53基因沒(méi)有突變歸納法(induction,統(tǒng)計(jì)上的可能):可能對(duì)我所遇見(jiàn)的每一個(gè)癌癥病人其p53基因都有突變,因此我歸納:假設(shè)p53基因突變則有癌癥假設(shè)法(abduction,科學(xué)上的創(chuàng)新):可能對(duì)某丙有癌癥,因此他的p53基因可能有突變;假設(shè)p53基因沒(méi)有突變,則某丙不會(huì)有癌癥Logic(3)HowtoMakeHypothesesIntroductionResearchLogicCausationHypothesisDiscussionKoch'sPostulates1.Theorganismmustbefoundinallanimalswiththedis.(butnotinhealthyanimals).2.Theorganismmustbeisolatedfromadis.animalandgrowninpureculture.3.Theculturedorganismshouldcausedis.whenintroducedintohealthyanimals.4.Theorganismmustbereisolatedfromtheexperimentallyinfectedanimal.因果相關(guān)

(Bradford-HillCriteria)一致性:所有的研究結(jié)果都一致強(qiáng)度特異性:沒(méi)有其他的原因或解釋能造成這樣的結(jié)果劑量-效應(yīng)性:因素愈強(qiáng),效果愈大時(shí)間性:原因一定要出現(xiàn)在結(jié)果之前生物學(xué)的可能性:生理病理學(xué)上的可能性凝聚性:與現(xiàn)有的理論不矛盾實(shí)驗(yàn)性:可用實(shí)驗(yàn)來(lái)改變?cè)蚨淖兘Y(jié)果HowtoMakeHypothesesIntroductionResearchLogicCausationHypothesisDiscussion科學(xué)研究的主要內(nèi)容變項(xiàng):被研究對(duì)象的某一屬性因時(shí)地人物不同而在質(zhì)或量上的變化操作型定義:根據(jù)可觀察測(cè)量或可操作的特徵來(lái)界定某一名詞或概念假設(shè):

變項(xiàng)之間的假設(shè)性關(guān)係假說(shuō):經(jīng)過(guò)證實(shí)的假設(shè)定律:經(jīng)過(guò)重複驗(yàn)證的假說(shuō)理論:有組織、有架構(gòu)、具有邏輯關(guān)係的假說(shuō)或定律醫(yī)學(xué)是不確定的科學(xué)也是可能性的藝術(shù)Medicineisascienceofuncertaintyandanartofprobability-WilliamOslerThemorepreciselythepositionisdetermined,thelesspreciselythemomentumisknowninthisinstant,andviceversa.

--Heisenberg,uncertaintypaper,1927科學(xué)是確定的學(xué)問(wèn)嗎?KarlRaimundPopper1902-1994真不能被證明，只有偽可以被證明(真理是到現(xiàn)在為止尚未被否証的假說(shuō))科學(xué)的假說(shuō)事實(shí)論點(diǎn)皆須客觀地驗(yàn)證，而非不證自明。不能完全依賴(lài)傳統(tǒng)、主觀信念、或常識(shí)來(lái)檢驗(yàn)科學(xué)知識(shí)?？茖W(xué)思維是要抱持懷疑、批判的態(tài)度。沒(méi)有不證自明(self-evidence)的事知識(shí)是來(lái)自人們的經(jīng)驗(yàn)所有知識(shí)是暫時(shí)性的強(qiáng)調(diào)科學(xué)知識(shí)應(yīng)基於經(jīng)驗(yàn)方面可觀察的假設(shè)之上?？茖W(xué)家目前採(cǎi)用的証據(jù)、方法和理論，來(lái)提出其暫時(shí)性的知識(shí)，它有可能隨時(shí)被修正。HypothesisAsuggestedexplanationforaphenomenonorofareasonedproposalsuggestingapossiblecorrelationbetweenmultiplephenomena.Thescientificmethodrequiresthatonecantestascientifichypothesis.Hypothesesarebasedonpreviousobservationsoronextensionsofscientifictheories.TypesofHypotheses1.AcausesB:Ifaparticularindependentvariablechanges,thenacertaindependentvariablealsochanges.Thisformulation,alsoknownasan"IfandThen"statement.2.AisrelatedtoB:Investigatorsmayhavemoredifficultyinverifyingcausalrelationshipsthanothercorrelations,becausequitecommonlyinterveningvariablesalsobecomeinvolved.假設(shè)(Hypothesis)有時(shí)候以問(wèn)題的方式提出,每一個(gè)假設(shè)都要有背景目的:解釋舊事實(shí)、預(yù)測(cè)新事實(shí)、挑戰(zhàn)舊理論、建構(gòu)新理論、尋找真理能答復(fù)的能被否證(falsifiable)的要素:研究的族群、變數(shù)(自變數(shù)與應(yīng)變數(shù))之間的關(guān)係(因果相關(guān)或純相關(guān))FunctionsofHypothesesToexplainoldfactsTopredictnewfactsTochallengeanoldtheoryTobuildanewtheoryTosearchfortruth(research=tosearchagain)HypothesesorQuestionsFalsifiable(measurable&testable):虛無(wú)假設(shè)(否証,假設(shè)P<0.05則否証失敗)、對(duì)立假設(shè)Answerable(methodstoanswer)Species/populationstudied,relationship(causalorcorrelational)betweenin-dependent&dependentvariablesHypothetico-deductivemodel1.Gatherdata(somethingthatisunknown,unexplained,ornew)2.Hypothesizeanexplanationforthoseobservations.3.Deduceaconsequenceofthatexplanation.(Aprediction)3.Designanexperimenttoseeifthepredictedconsequenceisobserved.4.Waitforcorroboration.Ifthereiscorroboration,gotostep3.Ifnot,thehypothesisisfalsified.Gotostep2.EvaluatingHypothesesTestibilitySimplicity(asintheapplicationof"Occam'sRazor",discouragingthepostulationofexcessivenumbersofentities)3.Scope-theapparentapplicationofthehypothesistomultiplecasesofphenomena4.Fruitfulness-theprospectthatahypothesismayexplainfurtherphenomenainthefuture5.Conservatism-thedegreeof"fit"withexistingrecognisedknowledge-systemsHypothesisApropositionaboutthesolutiontoaproblem,therelationshipofvariables:1.Anexpectedrelationshipbetweenvariables2.Basedoneithertheoryorevidence3.Testable4.Briefwithclarity5.Statedindeclarativeform6.BeoperationalbyeliminatingambiguityinthevariablesorproposedrelationshipsHypothesisTestibility(comparefalsifiabilityasdiscussedabove)Simplicity(asintheapplicationof"Occam'sRazor",discouragingthepostulationofexcessivenumbersofentities)Scope-theapparentapplicationofthehypothesistomultiplecasesofphenomenaFruitfulness-theprospectthatahypothesismayexplainfurtherphenomenainthefutureConservatism-thedegreeof"fit"withexistingrecognisedknowledge-systemsTestabilityofHypothesesContingency,defeasibility,orfalsifiability:CounterexamplestothehypothesisarelogicallypossibleFeasibilityofobservingareproducibleseriesofsuchcounterexamplesiftheydoexist.Inshort,ahypothesisistestableifthereissomerealhopeofdecidingwhetheritistrueorfalseofrealexperience.UponthispropertyofitsconstituenthypothesesreststheabilitytodecidewhetheratheorycanbeconfirmedorfalsifiedbythedataofactualexperienceFourComponentsofaHypothesis1.Subjectgroup:Whoareyouinterestedinstudying?2.Treatmentorexposure:Whatisbeingdonetoyoursubjectgroup?3.Outcomemeasure:Howisthetreatmentorexposureassessed?(definedpreciselyandunambiguously)4.Controlgroup:Whoareyoucomparingto.Itisimportantforthecontrolgrouptobeassimilaraspossibletothosewhoreceiveatreatmentorexposure.

Noteveryresearchhypotheseswillhaveallfourcomponents.Forexample,across-overdesigninvolvesapplyingbothanewtreatmentandastandardtreatmentusingthesamepatients.Forthisstudy,thehypothesiswouldnotinvolveaseparatecontrolgroup.Correlationalstudieslookatrelationshipswithinasinglegroup,suchasastudyofthefactorsthatcausemedicationerrors.Thistypeofstudywouldnothaveatreatment/exposure.Thestructurementionedhere,however,isstillusefulfordevelopingmostresearchhypotheses.統(tǒng)計(jì)的假設(shè)用反駁法來(lái)證明建立虛無(wú)假設(shè)〔nullhypothesis〕，以及替代假設(shè)(alternativehypothesis)虛無(wú)假設(shè)－自變數(shù)與應(yīng)變數(shù)之間沒(méi)有關(guān)聯(lián)替代性假設(shè)－自變數(shù)與應(yīng)變數(shù)之間有關(guān)聯(lián)以推翻虛無(wú)假設(shè)來(lái)支持實(shí)驗(yàn)的假設(shè)HowtoMakeHypothesesIntroductionResearchLogicCausationHypoth