Assertion Status

The clinical finding is explicitly stated as currently true for the patient. This is the default positive polarity indicating the condition or symptom is active.

• Linguistic cues: Direct statements without negation or hedging triggers
• Example: 'Patient has hypertension' or 'Type 2 diabetes is noted'
• Clinical significance: These findings populate the active problem list and drive billing codes
• System behavior: Entities classified as Present pass through to structured data extraction without inversion

The clinical finding is explicitly ruled out for the patient. The negation cue semantically reverses the existence of the concept within the document context.

• Key triggers: 'no', 'denies', 'without evidence of', 'negative for', 'ruled out'
• Example: 'Patient denies any chest pain' or 'No evidence of pulmonary embolism'
• Critical safety role: Prevents false attribution of diseases to the patient record
• Algorithmic approach: Negex and ConText use regular expressions to identify negation cues and determine their scope over target entities

The clinician expresses doubt or speculation about the finding without confirming or ruling it out. This represents epistemic modality where the author lacks full commitment to the proposition.

• Key triggers: 'possible', 'suspected', 'cannot rule out', 'suggestive of', 'likely', 'probable'
• Example: 'Suspected urinary tract infection' or 'Findings suggestive of interstitial lung disease'
• Downstream handling: Often routed for human review or flagged for follow-up diagnostic testing
• Detection method: ConText algorithm extends Negex to capture hypothetical and hedging language using lexical trigger lists

The finding is true but occurred in the past and is not active at the time of documentation. This context prevents outdated conditions from appearing on the current problem list.

• Key triggers: 'history of', 'prior', 'previous', 'resolved', 'no longer present'
• Example: 'History of myocardial infarction in 2019' or 'Prior cholecystectomy'
• Temporal distinction: Differs from negation because the event did occur, just not currently
• ConText classification: Handled as a distinct context separate from negation, using temporal trigger terms to assign the historical label

The finding is present but applies to someone other than the patient, typically a family member or contact. This prevents false attribution of conditions to the subject of the record.

• Key triggers: 'mother', 'father', 'sibling', 'family history of', 'contact with'
• Example: 'Patient's mother has breast cancer' or 'Family history significant for diabetes'
• Risk of misclassification: Without experiencer detection, family history conditions are incorrectly added to the patient's active diagnoses
• ConText extension: The algorithm identifies the experiencer slot to determine if the subject of the finding is the patient or another individual

The finding is discussed in a conditional or future-oriented context where it is not asserted as currently true. This captures speculative treatment plans and 'what-if' clinical reasoning.

• Key triggers: 'if', 'should', 'may need', 'will consider', 'in the event of'
• Example: 'If fever develops, will start antibiotics' or 'May need surgical intervention if conservative management fails'
• Distinction from uncertainty: Hypotheticals describe conditional future actions rather than current diagnostic speculation
• Scope challenge: The conditional scope often spans multiple sentences, requiring broader context windows than simple negation detection

The NLP task of classifying statements that express doubt, speculation, or hedging regarding a medical condition. This directly maps to the 'uncertain' or 'possible' assertion status.

• Detects uncertainty cues: 'possible', 'suspected', 'cannot rule out'
• Classifies epistemic modality—the linguistic expression of certainty degree
• Distinguishes tentative diagnoses from definitive assertions for accurate problem list generation

The binary or multi-class categorization of a clinical statement's factuality. Forms the output layer that assertion status labels populate.

• Positive polarity: affirmed, present findings
• Negative polarity: negated, absent findings
• Extended models include historical, hypothetical, and experiencer polarity classes
• Directly feeds structured data extraction pipelines and problem list reconciliation