-
Notifications
You must be signed in to change notification settings - Fork 11
Evidence Lines
- add more info here, including several examples showing cases where data is organized into evidence lines.
- use text from old Evidence Line page here
Evidence Lines are an independent, meaningful argument relevant to the validity of a target assertion. Evidence lines are comprised of one or more individual pieces of information that collectively contribute to a single argument, and in this capacity serve as 'evidence items'. These items can be varies pieces of data from a single study, or information from several lines of investigation that collectively support a single argument. To qualify as an evidence line, this argument must 'move the needle', but not necessarily 'tip the scales' - as often multiple lines of evidence are needed to confidently put forth a proposition as true.
In SEPIO-based models, evidence lines are organizing links between an Assertion and Evidence Items, wherein evidence items that contribute to the same independent argument for some target assertion are grouped under a single evidence line. SEPIO's organization around evidence lines is critical, as its allows us to capture exactly how information is used/interpreted as evidence:
- They organize evidence items according to arguments to which they contribute - i.e. what individual pieces of info collectively interpreted to make one argument for a target assertion.
- They allow us to capture the direction and strength of evidence provided for a particular assertion (which are attributes of the evidentiary relationship between the evidence items and the assertion, not attributes of the evidence items themselves).
- They allow us to capture provenance of the evidence interpretation process - who, when, and how an agent used information as evidence (e.g. if a particular set of rules such as the ACMG Guidelines was followed in this process).
The level of organization Evidence Lines provide is critical, as it is at the level of the discrete 'arguments' they represent, not at the level of individual Evidence Items, that we want to evaluate various aspects of the evidence. It is at this level that we want to describe the direction of evidence (does it support or refute), and the relative strength of the argument it makes in that direction (does it provide weak or strong support). It is at this level that we often want to quantify how many arguments exist and how diverse the evidence is. And it is at this level that we can understand why existing evidence may be insufficient or in conflict. Ultimately it is the explicit organization of evidence around Evidence Lines that enables the assessment of the overall quantity, quality, diversity, and concordance of evidence for a particular assertion - key features that support computational approaches for their evidence and provenance-based evaluation.
The ability to represent these details enables the assessment of the overall quantity, quality, diversity, and concordance of evidence for a particular assertion - key features that support computational approaches for their evidence and provenance-based evaluation.