Reference
Glossary
Key terms and definitions for clinical negligence professionals working with AI-powered medical record analysis.
Clinical Negligence
A legal claim arising when a healthcare professional provides treatment that falls below the accepted standard of care, causing harm to the patient. In the UK, clinical negligence (sometimes called medical negligence) requires establishing a duty of care, breach of that duty, and causation — that the breach caused or materially contributed to the patient's injury. Claims are typically assessed against the standard set by the Bolam test and refined by the Bolitho test.
Protocol Compliance
The degree to which a clinician's actions align with established clinical protocols and guidelines during the diagnosis, treatment, and management of a patient. In clinical negligence cases, protocol compliance is assessed by comparing the documented care against the relevant guidelines (such as NICE guidelines) that were in force at the time of treatment. Deviations from protocol are not automatically negligent, but unexplained departures that result in harm are a key indicator of breach of duty.
NICE Guidelines
Evidence-based recommendations published by the National Institute for Health and Care Excellence (NICE) that define the expected standard of care for specific conditions and clinical scenarios in England and Wales. NICE guidelines cover areas from cancer referral pathways to sepsis management, diabetes care, and mental health treatment. While not legally binding, courts frequently treat NICE guidelines as a benchmark for the standard of care a patient could reasonably expect.
Care Pathway
A structured, evidence-based plan that outlines the expected steps in the diagnosis, treatment, and ongoing management of a specific condition. Care pathways define what should happen, when it should happen, and who should be involved at each stage. In clinical negligence, deviations from the expected care pathway — such as skipped diagnostic steps, delayed referrals, or incorrect treatment sequences — are central to establishing whether care fell below an acceptable standard.
Breach of Duty
In clinical negligence law, breach of duty occurs when a healthcare professional's actions fall below the standard of care expected of a reasonably competent practitioner in the same field. The test for breach of duty in the UK is established by the Bolam test (1957): a doctor is not negligent if they acted in accordance with a practice accepted as proper by a responsible body of medical opinion. This was refined by the Bolitho test (1997), which requires that the accepted practice be logically defensible.
Causation
The legal requirement to prove that the breach of duty caused or materially contributed to the patient's injury. In clinical negligence, causation is often the most contested element. The claimant must demonstrate, on the balance of probabilities, that the harm would not have occurred (or would have been less severe) but for the negligent treatment. This requires analysis of what the outcome would have been had appropriate care been provided.
Severity Scoring
A systematic method of rating the seriousness of identified compliance deviations on a numerical scale. In AI-powered medical record analysis, findings are typically scored from 1 (minor deviation unlikely to affect outcome) to 10 (critical breach that likely caused or contributed to patient harm). Severity scoring helps solicitors and expert witnesses prioritise findings, focus resources on the most significant issues, and make informed decisions about case viability.
PII Sanitisation
The process of identifying and removing personally identifiable information (PII) from text before it is processed by an AI system. In the context of medical record analysis, PII sanitisation strips patient names, NHS numbers, dates of birth, addresses, GMC numbers, and dozens of other identifier types, replacing them with safe labelled placeholders (e.g., [PERSON], [NHS_NUMBER]). This ensures that no patient-identifiable data is sent to external AI services.
OCR (Optical Character Recognition)
Technology that extracts readable text from scanned images and documents. In medical record analysis, OCR is essential for processing paper-based records that have been scanned to PDF. Modern OCR systems can handle printed text, typed forms, and — with varying accuracy — handwritten clinical notes. Intelligent OCR detection analyses each page to determine whether OCR is needed, applying it only where the existing text extraction is insufficient.
RAG (Retrieval-Augmented Generation)
An AI architecture that combines information retrieval with text generation to produce answers grounded in specific source documents. In medical record analysis, RAG enables AI case chat: when a user asks a question, the system retrieves the most relevant sections of the medical record using semantic search, then generates an answer based specifically on that retrieved context — with page-level citations. This approach prevents hallucination by grounding every response in the actual document.
Semantic Search
A search technique that matches queries based on meaning rather than exact keyword matches. In protocol compliance analysis, semantic search converts both the medical record text and clinical protocol sections into mathematical representations (vector embeddings) that capture their meaning. This allows the system to find relevant protocol guidance even when the medical record uses different terminology — for example, matching "chest pain on exertion" with a protocol section about "angina pectoris".
AES-256-GCM Encryption
An advanced encryption standard used to protect sensitive data at rest. AES-256-GCM (Advanced Encryption Standard with 256-bit keys in Galois/Counter Mode) is the same encryption standard used by government agencies and financial institutions. It provides both confidentiality (data cannot be read without the key) and integrity (any tampering with encrypted data is detected). In medical record analysis platforms, it protects all stored text, findings, and chat data.
Vector Embedding
A mathematical representation of text that captures its semantic meaning as a series of numbers (typically 1,536 dimensions). In medical record analysis, each section of text is converted into a vector embedding, enabling semantic search — the ability to find related content based on meaning rather than keyword matching. This is the foundation of both the AI case chat (RAG) and the protocol compliance analysis features.
GDPR (General Data Protection Regulation)
The primary data protection legislation governing the processing of personal data in the UK (retained as UK GDPR after Brexit). For medical record analysis, GDPR is particularly significant because medical records contain special category data (health data) under Article 9, which requires additional safeguards. Key GDPR requirements include having a lawful basis for processing, data minimisation, data retention limits, the right to erasure, and comprehensive audit logging.
Medical Chronology
A structured timeline of all significant clinical events in a patient's medical history, organised chronologically. In clinical negligence case preparation, building a medical chronology is one of the first and most important steps — it establishes when events occurred, what decisions were made, and where gaps or delays exist. AI tools can automate chronology extraction by identifying and categorising events such as admissions, diagnoses, treatments, medications, test results, and referrals.
ICO (Information Commissioner's Office)
The UK's independent authority responsible for upholding information rights and data protection. Organisations that process personal data in the UK must register with the ICO. For AI platforms handling medical records, ICO registration and compliance with ICO guidance on automated decision-making, data protection impact assessments, and special category data processing are essential requirements.
See MedCase AI in Action
Book a demo to see how AI-powered compliance analysis works for clinical negligence cases.