IMED2003 - Intro to Biomedical Research (L1 + L2)

What features define good experimental design?

Validity: measurements or conclusions align with the real world.

Reliability: repeating the same experiment many times gives similar results.

Reproducibility: another researcher following the same methods obtains similar results.

Good design also removes or accounts for confounding variables.

How do replicates and variable control improve biological experiments?

Biology is highly variable and complex.

Replicates reduce the effect of natural variability on results.

Controlling variables reduces variability and promotes reproducibility.

Researchers should identify other factors that could influence the result, then limit or account for their effects.

Lecturer explanation:

Biological variability can be seen even among pigeons, cabbages and identical twins.

What did the European scientific revolution change about knowledge of the natural world?

Before the scientific revolution:

- Knowledge production was exclusive and often required authority or religious sponsorship.

- Knowledge was controlled because few people outside the church could read.

- New knowledge was mainly derived deductively from existing facts.

After the European scientific revolution in the 16th century:

- Knowledge became more transparent: its derivation was explained.

- Claims became verifiable and open to being understood or refuted by anyone able to reason.

- Science became empirical, focusing on acquiring and interpreting new evidence.

- Deductive and inductive reasoning were balanced, with greater value placed on keeping an open mind.

What does the word "research" imply about the goal of inquiry?

The word derives from terms meaning to examine closely, seek, look for, go about, wander or traverse.

Research is therefore a process of searching and examining in order to get closer to the truth.

How do opinions and arguments differ?

Opinions are beliefs that do not require justification.

Arguments are justified beliefs: claims supported by evidence.

Understanding what makes a good argument allows scientific claims to be evaluated.

What are the main components of an argument?

Conclusion: the claim or assertion.

Premises: established facts or supporting statements.

Reasoning: the logical connection showing how the premises support the conclusion.

Sub-arguments: arguments that support a premise used in another argument.

What three criteria make an argument good?

1. The premises are acceptable.

2. The premises are relevant to the conclusion.

3. The premises provide adequate grounds for supporting the conclusion.

Mnemonic used in the lecture: ARG — Acceptable, Relevant, Grounds.

What is deductive entailment?

Deductive entailment is the strongest form of logical support.

If statement A deductively entails statement B, it is logically impossible for A to be true while B is false.

A deductively valid argument has relevant premises that provide grounds for the conclusion, although it can still fail if its premises are unacceptable.

What categorical form illustrates a deductively valid argument?

Premise 1: All x are y.

Premise 2: z is an x.

Conclusion: z is a y.

Lecture example:

Premise 1: All mammals are animals that have live young.

Premise 2: Cats are mammals.

Conclusion: Cats are animals that have live young.

What qualifies an observation statement as a scientific fact?

An observation statement can form part of the basis of science when it can be straightforwardly tested by the senses and withstands those tests.

Science derives knowledge from facts of experience through reasoning.

What supports scientific assertions about the natural world?

Scientific assertions are supported by cogent arguments containing:

- Verifiable evidence.

- A sound logical link between premises and conclusion.

- A conclusion whose strength is proportional to the strength of the evidence.

This produces rational belief rather than absolute certainty.

Why does observing many white swans not prove that all swans are white?

No finite number of white-swan observations can establish the universal claim "All swans are white".

The claim cannot be empirically verified because observing every swan is impossible.

One black swan is sufficient to falsify the universal claim.

Therefore, the observations may be acceptable and relevant but do not provide adequate grounds for certainty.

How should the scope and commitment of a scientific conclusion relate to its evidence?

Conclusions with a large scope require premises with a similarly large scope, but true large-scope premises are difficult to obtain.

Scientists can:

- Narrow the scope of a conclusion.

- Express less commitment to it.

- Match confidence in the claim to the available evidence and the cogency of the argument.

What is inductive support?

Inductive support uses observations of particular cases to infer a claim about a larger group, future event or unobserved situation.

It includes generalisations, predictions and explanations.

It assumes that the world and our experience of it are sufficiently regular and consistent.

Even the strongest inductive argument cannot prove its conclusion with certainty.

What is inference to the best explanation?

Premise 1: An observation.

Premise 2: A hypothesis proposed as the best and most probable explanation of that observation.

Conclusion: Most probably, the hypothesis is true.

The hypothesis supplies the grounds linking the observation to the conclusion.

How can the relevance of premises in an inductive scientific argument be improved?

Use cases that are representative of the target population in all relevant respects.

Because representativeness cannot be guaranteed:

- Use statistical methods to guide selection, including sample size.

- Use good study design to improve sampling and reduce bias.

- Construct an explicit argument for why the sample represents the population.

How are statistics incorporated into an inductive scientific argument?

Premise 1: Observations from groups with and without the proposed causal factor.

Premise 2: The hypothesis is the best and most probable explanation.

Statistics premise: Quantifies how strongly the evidence supports the hypothesis.

Conclusion: A stated level of confidence that the hypothesis is true.

Statistics help match the scope of the evidence and the degree of commitment to the claim.

What are the three interacting components of science?

Scientific method:

- Empirical observation

- Hypothesising

- Testing

Scientific knowledge:

- A collection of tested theories and facts

Scientific community:

- Challenges, validates and organises scientific work

All are centred on transparency, verifiability and openness.

DIAGRAM ON SLIDE 26

What are the classic stages of the scientific method?

1. Observe: make observations.

2. Question: identify a problem or ask a question.

3. Research: search for existing answers or solutions.

4. Hypothesise: formulate a hypothesis.

5. Experiment: design and perform a test; collect and analyse data.

6. Test the hypothesis: accept or reject it based on evidence.

7. Draw conclusions based on the test.

8. Report and share the results.

DIAGRAM ON SLIDE 29

How do observations, hypotheses and experiments contribute to scientific knowledge?

Observations generate hypotheses.

A hypothesis is a provisional explanation or prediction.

Experiments or other validation methods test the hypothesis.

A supported or verified hypothesis contributes to scientific knowledge.

What makes a scientific hypothesis well formulated?

It is a precisely formulated scientific question.

It is usually expressed conditionally in an "if ... then ..." form.

It must be falsifiable so evidence can lead to its rejection or acceptance.

"Is there a difference if ...?" is too vague unless made precise.

Why must experiments control other potentially causal factors?

Observed effects can result from several causes and processes.

Methodology must therefore eliminate or account for competing causal factors:

- Experimentally, through controls.

- Analytically, through data analysis.

This allows the resulting observations to test the intended hypothesis more directly.

What resources and processes contribute to research design?

Research design requires:

- Expertise

- Technology

- Work

- Trial and error

These are used to create observations capable of testing a hypothesis while controlling other relevant factors.

Why is the scientific method iterative rather than simply linear?

Investigations are refined over time.

New technologies permit new kinds of observations and greater measurement accuracy.

New knowledge and technology can reveal problems in earlier experiments.

New data may conflict with current understanding, leading to revised hypotheses, new tests and updated theories.

How do scientific theories influence observations and hypotheses?

Theories are distilled from tested hypotheses, but they also shape future science.

They guide:

- Which observations scientists consider relevant.

- Which hypotheses are plausible and worth testing.

- How findings are interpreted within existing knowledge.

How is the real scientific process broader than the classic linear model?

Science is non-linear, iterative, complex, organic and collaborative.

Observations, questions, hypotheses, experiments, theory, communication and community feedback influence one another rather than occurring in one fixed sequence.

DIAGRAM ON SLIDE 36

How is scientific knowledge revised when new data conflict with current understanding?

1. Make the new observations.

2. Formulate a new or revised hypothesis that explains them.

3. Test whether the hypothesis is supported.

4. Integrate supported findings into theories.

5. Review and revise scientific knowledge to improve understanding of the world.

What are the defining attributes of a scientific theory?

A scientific theory:

- Synthesises many independent investigations.

- Is based on data but is more than a collection of data.

- Has been widely tested.

- Is accepted as valid and accurate in proportion to the evidence.

- Has explanatory and predictive power.

What are explanatory and predictive power in a scientific theory?

Explanatory power: existing evidence is consistent with the theory and can be explained by it.

Predictive power: the theory can be used to make accurate predictions about new observations.

Theories are tested by evaluating both forms of power.

What conceptual functions do scientific theories serve?

Theories provide structured "ways of thinking".

They define boundaries around:

- What is relevant to the theory.

- Where and how the theory can appropriately be applied.

What do scientific journal articles record?

Journal articles serve as records of the scientific method, including:

- Observations

- Hypotheses

- Experiments or other tests

- Results

- Conclusions

What role does the introduction of a research paper play?

It outlines the current evidence and relevant theories.

It shows why the question is not trivially answered.

It identifies the research gap that the study is intended to address.

What does a scientific journal editor do after receiving a manuscript?

The editor assesses whether the work addresses a question relevant to the journal's aims and scope.

Selected submissions are then sent for peer review.

Journal scope helps define and maintain the boundaries of scientific fields and theories.

How did peer review develop historically?

Early scientific journals were linked to scientific societies and lacked consistent standards; access could be exclusive or censored.

Around the 1750s, editorial committees began referring work to society fellows for expert review.

By about 1830, submissions required written reports on their fitness for publication.

Nature was founded in 1869 and adopted peer review as standard practice in 1973 to reduce accusations of cronyism and elitism.

Peer review is now a defining feature of scientific research.

What normally happens during peer review?

1. An editor sends the article to usually three peer reviewers who are peers of the authors.

2. Each reviewer recommends:

- Accept

- Reject and resubmit after additional work

- Reject

3. Accepted papers are published.

4. Rejected papers may be revised or submitted to another journal.

What are the main limitations of peer review?

Reviewers are generally unpaid.

Reviewers may fail to identify research problems.

Published papers vary in quality.

Impact factor is sometimes used as a broad proxy for quality because influential work may be cited more often, but it is not a guarantee.

The reader must still evaluate the quality of published research.

How does citation facilitate scientific research?

Published research becomes part of the scientific literature.

Theories synthesise many investigations.

Current studies build on past findings and theories.

Citation makes the supporting premises transparent and helps readers assess whether they are acceptable, relevant and sufficient to support the conclusion.

Why is science described as getting closer to the truth rather than proving absolute truth?

Claims derived from experience cannot be deductively proved with certainty.

Science therefore:

- Forms rational arguments.

- Makes commitment proportional to evidence.

- Refutes or revises theories when contradicted.

- Treats current knowledge as the best available understanding at the present time.

This explains why today's best evidence may later be replaced.

Why is understanding the epistemological structure of science important in healthcare?

Healthcare built on scientific knowledge aims to be effective, ethical and consistent rather than based on unsupported "miracle cures".

Understanding how scientific claims are constructed helps clinicians evaluate whether apparently scientific claims are genuinely supported.

Why are scientific theories clinically useful?

Their explanatory power helps explain disease mechanisms.

Their predictive power supports forecasts about diagnosis, prevention and treatment.

However, theoretical plausibility must still be tested empirically before changing clinical practice.

How did cervical cancer screening illustrate the use of theory and empirical research?

Mechanistic understanding:

- Most cervical cancer results from human papillomavirus (HPV) infection.

- Viruses contain DNA.

- PCR can detect specific DNA sequences with high sensitivity and specificity.

Clinical development:

- From the 1980s to 2017, Pap smears screened for precancerous cellular changes every 2 years.

- After 2017, HPV PCR testing was used every 5 years.

Empirical research was still required to determine whether HPV PCR screening was effective and how often it should be performed.

How can extrapolating from mechanistic understanding mislead clinical practice?

After myocardial infarction, damaged heart tissue led to the theoretical advice that complete bed rest would aid recovery.

Empirical research showed that bed rest increases thromboembolism risk.

Advice was revised to balance rest with monitored exercise.

Mechanistic reasoning is useful but cannot replace outcome evidence.

What does the statement "Half of what is taught in medical education is wrong, but we don't know which half" illustrate?

Medical knowledge is provisional.

Clinicians must be prepared for current teaching to be revised as stronger evidence emerges.

The statement is attributed to Dr C. Sidney Burwell and was cited in a 1956 BMJ article.

How did the interpretation of "junk DNA" demonstrate that scientific knowledge changes?

A 2003 molecular cell biology textbook described about 40% of the human genome as having unknown function, and some called it "junk DNA".

About 13 years later, the same textbook described much more of its structure and function.

New technology and research had refined the earlier understanding.

Why does scientific knowledge change over time?

- New research improves understanding.

- Better technologies enable new discoveries and more accurate observations.

- New treatments are discovered and developed.

- Biases in earlier study designs are recognised.

- Theoretical understanding of health and disease is refined.

What skills are required to work with changing scientific evidence in professional practice?

1. A sound understanding of the current theoretical knowledge base.

2. The ability to evaluate and interpret new scientific findings.

3. The ability to incorporate valid and applicable new knowledge into practice.

What criteria should be used to evaluate whether research is worth putting into practice?

Novelty and relevance:

- Answers a genuine research question.

- Engages with the existing literature.

- Identifies a research gap.

Internal validity:

- Methodologically sound.

- Accurate data collection.

- Appropriate controls.

- Reduced bias and confounding.

- Appropriate sample.

Applicability or external validity:

- Relevant and generalisable to the population of interest.

Impact:

- Clinically significant and useful.

- Produces a meaningful effect if implemented.

What characteristics define high-quality research?

- Identifies and addresses a research gap.

- Uses a study design appropriate to the research question.

- Minimises bias and confounding.

- Applies techniques, including statistics, appropriately.

- Produces consistent and reproducible findings.

- Interprets results logically.

- Meets ethical requirements, including appropriate funding disclosure and treatment of humans or animals.

What are the major types of biomedical research questions?

1. Treatment or therapy

2. Prevention

3. Diagnosis and screening

4. Prognosis

5. Aetiology or causation

6. Meaning or patient experience

7. Mechanistic questions in basic or bench science

DIAGRAM ON SLIDE 68

What does a treatment or therapy research question ask?

It asks whether an intervention improves outcomes in people who already have a condition.

Interventions may include medication, surgery, exercise or counselling.

What does a prevention research question ask?

It asks whether an intervention or exposure prevents morbidity or mortality in people who do not yet have the disease.

Both benefits and possible harms must be evaluated.

The practical aim is to reduce the chance of disease occurring.

What does a diagnosis or screening research question ask?

It asks how accurately a test or procedure distinguishes people with a condition from those without it.

It can address:

- Which test is appropriate.

- How diagnostic results should be interpreted.

What does a prognosis research question ask?

It concerns the expected course of a disease, illness or condition.

Forms include:

- Overall prognosis: incidence of an outcome over a stated time in people with a health state.

- Prognostic factors: features that predict outcomes.

- Prognostic models: combinations of factors used to predict outcomes in individuals.

- Treatment selection: factors or models that predict response to particular treatments.

What does an aetiology or causation research question ask?

It investigates possible causes of disease or factors that increase or decrease disease risk.

It can be viewed as the inverse of a therapy question because it considers harmful outcomes of an exposure.

Examples:

- What harm is caused by an exposure?

- What causes a disease or makes it more likely?

What does a meaning research question ask?

It explores why things happen from the patient's perspective.

It focuses on attitudes, experiences, beliefs, opinions and perceptions.

It seeks to understand behaviour and what a patient's experience of a situation may be.

These questions commonly use qualitative research.

Why are standard study designs useful in biomedical research?

They promote consistency in experimental design.

They make research easier to perform, compare and evaluate.

The preferred design depends on the type of research question.

Which study designs are preferred for each major clinical question type?

Intervention, therapy or prevention:

RCT > cohort study > case-control study > case series

Diagnosis:

Prospective, blinded comparison with a gold standard

Aetiology or causation:

RCT > cohort study > case-control study > case series

Prognosis:

Cohort study > case-control study > case series

DIAGRAM ON SLIDE 73

What is evidence-based practice?

Evidence-based practice is the conscientious, explicit and judicious use of current best evidence in decisions about individual patient care.

It integrates:

- Best available research evidence

- Clinical expertise

- Patient context and values

- Practice context

DIAGRAM ON SLIDE 75

What two fundamental principles underpin evidence-based practice?

1. Not all evidence is equal:

Research quality must be evaluated, and hierarchies of evidence help distinguish information more likely to be valid.

2. Evidence alone is insufficient:

Clinical decisions must also incorporate clinical expertise, patient values and preferences, and the practice context.

Why should health professionals engage with evidence-based practice?

- To optimise patient outcomes.

- To make the best use of resources.

- To promote an attitude of inquiry: "Why do I do things this way?"

- To encourage self-reflection.

- To maintain professional accountability by avoiding outdated techniques.

How can evidence-based practice improve both clinical practice and research?

It creates a reciprocal relationship:

- Better research improves professional decisions.

- Critical use of evidence exposes gaps and weaknesses in research.

- Those identified problems guide the development of stronger future studies.

What are the five steps in the 5 A's model of evidence-based practice?

1. Assess: a question arises during assessment of a patient or patient group.

2. Ask: completely articulate the clinical question.

3. Acquire: conduct a focused search and select the highest-quality evidence.

4. Appraise: evaluate validity and clinical applicability.

5. Apply: integrate applicable evidence with clinical expertise and patient values.

What should be assessed before constructing a clinical question?

- Perform a clinical evaluation.

- Take a thorough patient history.

- Conduct a physical examination.

- Understand the relevant pathophysiology.

These details supply the information required to build a focused PICO question.

What makes a good clinical question?

It:

- Focuses on one or two key issues.

- Is directly relevant to the patient or problem.

- Is phrased to facilitate a literature search.

- Produces an answer that addresses the clinical scenario.

A framework helps ensure all important factors are considered.

What does PICO stand for?

P — Patient, population or problem of interest

I — Intervention, exposure, diagnostic test, prognostic factor or treatment

C — Comparison or control

O — Outcome

What details should be considered for each PICO element?

P:

- Patient demographics such as age, sex or ethnicity

- The problem or condition

I:

- The intervention being considered, such as medication, exercise, rest, exposure or diagnostic test

C:

- Another treatment, placebo, usual care or no treatment

O:

- Desired benefits

- Undesirable effects

- Side effects or harms

How was PICO applied to the work-related neck pain example?

P — Work-related neck muscle pain

I — Strength training

C — Rest

O — Reduction in pain

What does PICO(TT) add to the PICO framework?

The first T identifies the type of evidence.

The second T identifies the type of research design.

These additions help locate the strongest appropriate evidence for the type of clinical question.

How can PICO(TT) be used to judge whether a publication is relevant to a clinical question?

Compare the publication's:

- Patient or population

- Intervention or exposure

- Comparison

- Outcome

- Type of evidence

- Research design

with those of the clinical question.

Close alignment strengthens the relevance of the evidence and the grounds for applying its conclusion.

What occurs during the "Acquire" step of evidence-based practice?

Conduct a thorough, focused search guided by the clinical question.

Select the highest-quality evidence that is relevant to the PICO(TT) elements and question type.

What occurs during the "Appraise" step of evidence-based practice?

Evaluate the evidence for:

- Validity

- Methodological quality

- Clinical relevance

- Generalisability

- Applicability to the patient and practice context

What five outcomes are assessed in an evidence-grading system?

1. Evidence base:

Types and sizes of studies, such as case reports, cohorts, RCTs and meta-analyses.

2. Consistency:

Whether the body of literature agrees.

3. Clinical impact:

The effect on morbidity or other meaningful outcomes.

4. Generalisability:

Whether findings apply to the target population, including Australia.

5. Applicability:

Whether the findings should be implemented in the relevant guideline or setting.

What did the Australian Dietary Guidelines evidence example illustrate?

A large evidence report can synthesise enormous amounts of research: the example was an 1105-page report drawing on about 55,000 peer-reviewed publications.

Evidence statements were graded by evidence base, consistency, clinical impact, generalisability and applicability.

DIAGRAM ON SLIDE 91

What conclusion was shown in the evidence table about long-chain PUFA and all-cause cancer?

Evidence statement:

Consumption of long-chain polyunsaturated fatty acids was not associated with total all-cause cancer incidence or mortality.

Overall grade:

C

Ratings:

- Evidence base: Good

- Consistency: Good

- Clinical impact: Poor

- Generalisability: Good

- Applicability: Satisfactory

The table noted broad adult populations but cautioned that much evidence came from the United States and Europe, where dietary patterns may differ from Australia.

DIAGRAM ON SLIDE 92

What features and consequences characterise low-quality research?

Possible problems:

- Bias or confounding caused by poor study design.

- Interpretation not supported by the data.

Consequences:

- Distorted results.

- Reduced confidence about what will happen if the findings are applied in practice.

What current concerns arise from using generative AI in biomedical research?

Defining acceptable evidence:

Standards must be maintained when evaluating biomedical applications of generative AI.

Maintaining scientific rigour:

A literature flooded with spurious or misleading findings could compromise evidence-based practice.

What central standard should be applied to claims about medical AI?

Claims that medical AI improves care must be supported by appropriate evidence.

The scientific community should demand proportionate validation rather than accepting theoretical promise alone.

Lecturer explanation:

The lecture connected this issue to the scientific values of transparency, verifiability and openness.

Why can misuse of AI-assisted research tools threaten the scientific literature?

Inexperienced or inappropriate use can:

- Produce misleading analyses.

- Miss bias or confounding.

- Generate publishable-looking but unreliable findings.

- Encourage selective reporting or "pick-positive-results" practices.

- Add spurious papers to the evidence base.

This can mislead clinicians and weaken evidence-based practice.

What broader criticism was made about AI-generated scientific publishing?

Rigorous checking of AI-generated papers can consume substantial human time.

Rather than automatically making publishing faster or cheaper, poorly used AI may make the process slower, worse and more expensive.

Human judgement remains necessary to verify evidence and reasoning.

What four domains of capability are needed to work responsibly with medical AI?

1. Technical concepts

2. Validation

3. Ethics

4. Appraisal

These capabilities support different roles, including consumers, translators and developers of healthcare AI.

DIAGRAM ON SLIDE 100