Hover (or tap) a card to reveal the answer. Filter by topic below.
A type of machine learning where a model is trained on labelled input–output pairs. The model learns a mapping function and generalises to unseen data. Common algorithms: logistic regression, random forests, gradient boosting (LightGBM, XGBoost), and neural networks.
Machine learning on unlabelled data — algorithms discover hidden structure or groupings. Key methods: k-means, HDBSCAN (density-based clustering), GMM (Gaussian Mixture Models), and PCA / UMAP for dimensionality reduction.
Partitions data into k equal folds. The model trains on k−1 folds and validates on the remaining fold, rotating until every fold has been the validation set. Reduces variance in performance estimates.
Using domain knowledge to create, transform, or select input variables that improve model performance. Includes encoding categoricals, normalisation, log transforms, interaction terms, and lag features.
Extract, Transform, Load. A data engineering workflow that pulls raw data from source systems, applies cleaning and transformation logic, and loads it into a downstream store such as a data warehouse or ML feature store.
Identifying data points that deviate significantly from expected behaviour. Used in quality control, fraud detection, and clinical outlier flagging. Methods: Isolation Forest, Z-score thresholding, autoencoder reconstruction error.
High-throughput DNA/RNA sequencing technologies that parallelise sequencing reactions to produce millions of short reads simultaneously. Enables whole-genome sequencing, RNA-seq, and targeted panel sequencing at scale.
Statistical identification of genes whose expression levels differ significantly between conditions. Key outputs: log2 fold-change and adjusted p-value (FDR). Tools: DESeq2 (negative binomial model) and EdgeR.
Identifying genetic variants (SNPs, indels, structural variants) from sequencing reads by comparing to a reference genome. Common tools: GATK HaplotypeCaller, Mutect2. Output format: VCF (Variant Call Format).
Standard raw sequencing data format. Each read has 4 lines: (1) sequence identifier, (2) nucleotide sequence, (3) "+" separator, (4) Phred quality scores as ASCII characters indicating base-call confidence.
Mapping sequencing reads to a reference genome to determine their genomic origin. Splicing-aware aligners (STAR, HISAT2) are used for RNA-seq. Output: BAM/SAM file with read positions, CIGAR strings, and mapping quality.
A controlled vocabulary describing gene product functions across three domains: Biological Process, Molecular Function, and Cellular Component. Used in enrichment analysis to interpret differentially expressed gene sets.
A tumour-specific antigen arising from somatic mutations that alter a protein's amino acid sequence. Recognised as foreign by the immune system — key targets in personalised cancer immunotherapy and vaccine development.
A workflow orchestration framework for scalable and reproducible bioinformatics pipelines. Supports parallel execution, containerisation (Docker/Singularity), and deployment across HPC and cloud environments (GCP, AWS).
Condition that comes on quickly and often lasts a short time — e.g. heart attack, the flu, or an asthma attack.
The right of patients to make decisions about their own healthcare.
Persisting for a long time or recurring — e.g. high blood pressure, diabetes, or arthritis.
The ability to understand and share another person's feelings and experience.
Individuals having enough knowledge, understanding, skills and confidence to use health information. Affects ability to engage with self-care, navigate the healthcare system, interpret risk, and evaluate online information quality.
Considering all physical, psychological and social factors potentially impacting a person's health.
Sorting patients based on urgency of need, prioritising those requiring the most immediate treatment.
Assesses an action by its consequences — the net benefits and costs to all stakeholders. Strives to achieve the greatest good for the greatest number while minimising harm.
Organisational level between cells and organ — an ensemble of similar cells and their extracellular matrix that carry out a specific function.
Cells that can differentiate into other cell types (multipotent) and divide by self-renewal to produce more stem cells.
A disease of muscle in which muscle fibres do not function properly.
A molecule with both hydrophilic and hydrophobic parts. Phospholipids in cell membranes are a classic example — they self-assemble into bilayers with tails inward and heads outward.
Protecting peoples' health, wellbeing and human rights, enabling them to live free from harm, abuse and neglect.
A serious condition where the heart cannot pump enough blood and oxygen to vital organs. Most commonly caused by a heart attack. A medical emergency requiring immediate treatment.
Life-threatening condition from losing more than 20% of blood or fluid — the heart cannot pump sufficient blood, risking organ failure.
A severe, potentially life-threatening allergic reaction causing blood pressure to drop so low that cells and organs are deprived of oxygen.
Chest pain caused by reduced blood flow to the heart — a symptom of coronary artery disease. Described as squeezing, pressure, or tightness in the chest. Medication includes Glyceryl trinitrate (GTN) spray.
Occurs when blood flow decreases or stops to part of the heart muscle. Common symptoms: chest pain radiating to shoulder, arm, back, neck or jaw. Can lead to cardiac arrest.
Bluish discolouration of skin or lips caused by low oxygen levels in red blood cells or problems delivering oxygenated blood to the body.
Also called coronary heart disease (CHD) — heart problems caused by narrowed coronary arteries reducing blood and oxygen supply to the heart muscle.
Biomedical discipline integrating biology with engineering to create tissues or cellular products outside the body (ex vivo) or to support repair within the body (in vivo).
A push or pull upon an object resulting from its interaction with another object. Measured in Newtons (N).
The degree of contact match between two joint surfaces. High congruence = surfaces match closely. Low congruence increases stress concentrations at contact points.
The science and engineering of interacting surfaces in relative motion — encompassing friction, wear, and lubrication. Critical in joint replacement design.
Bones adapt to the mechanical demands placed on them — increased load leads to remodelling and strengthening; reduced load leads to bone loss.
A measure of material stiffness — the ratio of stress (force per unit area) to strain (proportional deformation). Higher Young's modulus = stiffer material.
The design of materials, structures and systems modelled on biological organisms and processes — e.g. gecko-inspired adhesives, shark-skin drag-reduction surfaces.
An artificial substitute for a missing body part (e.g. eye, limb, or tooth), used for functional or cosmetic purposes, or both.
The quality of being toxic to cells. A key consideration when evaluating the biocompatibility of implant materials.
Scattering of light by particles much smaller than the wavelength of light, without change in wavelength. Accounts for the blue colour of the sky — blue light scatters more efficiently than red.
The emission of electrons (photoelectrons) when electromagnetic radiation hits a material. Underpins many biomedical imaging and sensing technologies.