Temporal Reasoning and Planning in Medicine Temporal Mediators

  • Slides: 32
Download presentation
Temporal Reasoning and Planning in Medicine Temporal Mediators: Integration of Temporal Reasoning and Temporal-Data

Temporal Reasoning and Planning in Medicine Temporal Mediators: Integration of Temporal Reasoning and Temporal-Data Maintenance Yuval Shahar MD, Ph. D

Temporal Reasoning and Temporal Maintenance n n n Temporal reasoning supports inference tasks involving

Temporal Reasoning and Temporal Maintenance n n n Temporal reasoning supports inference tasks involving time-oriented data; often connected with artificial-intelligence methods Temporal data maintenance deals with storage and retrieval of data that has multiple temporal dimensions; often connected with database systems Both require temporal data modelling

Examples of Temporal. Maintenance Systems TSQL 2, a bitemporal-database query language (Snodgrass et al.

Examples of Temporal. Maintenance Systems TSQL 2, a bitemporal-database query language (Snodgrass et al. , Arizona) n TNET and the TQuery language (Kahn, Stanford/UCSF) n The Chronus/Chronus 2 projects (Stanford) n

Examples of Temporal-Reasoning Systems n RÉSUMÉ M-HTP n TOPAZ n Tren. Dx n

Examples of Temporal-Reasoning Systems n RÉSUMÉ M-HTP n TOPAZ n Tren. Dx n

A Typical TM and TR Application: Automated Support to Therapy by Clinical Guidelines/Protocols §

A Typical TM and TR Application: Automated Support to Therapy by Clinical Guidelines/Protocols § Clinical guidelines/protocols contain recommendations for medical interventions that are predicated on the observation of: u relevant temporal patterns of these states u relevant patient states

CCTG-522 Recommendation Modify the standard dose of AZT for a patient if, during treatment

CCTG-522 Recommendation Modify the standard dose of AZT for a patient if, during treatment with the protocol, the patient experiences a second episode of moderate anemia that has persisted for more than two weeks

Protocol-Based Decision Support System n n Presents patient-specific recommendations Needs a method for verifying

Protocol-Based Decision Support System n n Presents patient-specific recommendations Needs a method for verifying the presence of timeoriented patient conditions in a database

Information Mismatch

Information Mismatch

Temporal Abstraction n Defined as the creation of high-level summaries of time-oriented data n

Temporal Abstraction n Defined as the creation of high-level summaries of time-oriented data n Necessary because u clinical databases usually store raw, time-stamped data u protocols often require information in high-level terms

Temporal Patterns

Temporal Patterns

Temporal Maintenance n Defined as the storage of time-oriented data and the selective retrieval

Temporal Maintenance n Defined as the storage of time-oriented data and the selective retrieval of that data based on some time-oriented constraint n Necessary because clinical conditions may be defined as temporal patterns u temporal order u temporal duration u temporal context

Temporal Data Manager n Performs u temporal abstraction of time-oriented data u temporal maintenance

Temporal Data Manager n Performs u temporal abstraction of time-oriented data u temporal maintenance n Used for tasks such as finding in a patient database which patients fulfil eligibility conditions (expressed as temporal patterns), assessing the quality of care by comparison to predefined timeoriented goals, or visualization temporal patterns in the patient data

Embedding A Temporal Data Manager Within a Guideline-Support System n Can be embedded within

Embedding A Temporal Data Manager Within a Guideline-Support System n Can be embedded within a larger decision support framework, e. g. , EON n Mediates all access to the external clinical database

Two Implementation Strategies 1) Extend DBMS 2) Extend Application

Two Implementation Strategies 1) Extend DBMS 2) Extend Application

Problems Extending DBMS Temporal data management methods implemented in DBMS: u are limited to

Problems Extending DBMS Temporal data management methods implemented in DBMS: u are limited to producing very simple abstractions u are often databasespecific

Problems Extending Applications Temporal data management methods implemented in applications: u duplicate some functions

Problems Extending Applications Temporal data management methods implemented in applications: u duplicate some functions of the DBMS u are application-specific

Our Strategy n Separates data management methods from the application and the database n

Our Strategy n Separates data management methods from the application and the database n Decomposes temporal data management into two general tasks: u temporal abstraction u temporal maintenance

The Tzolkin Temporal Mediator Architecture

The Tzolkin Temporal Mediator Architecture

RÉSUMÉ: Temporal Abstraction n Creates summaries of time-oriented data u Clinical data is usually

RÉSUMÉ: Temporal Abstraction n Creates summaries of time-oriented data u Clinical data is usually stored as “low-level” data u Protocols often specify conditions as “high-level”, interval-based concepts n n Is domain-independent Has a tool that facilitates knowledge acquisition and maintenance

Temporal Abstraction of Hb

Temporal Abstraction of Hb

Chronus: Temporal Maintenance n Provides temporal extensions to SQL n Historical relational model u

Chronus: Temporal Maintenance n Provides temporal extensions to SQL n Historical relational model u Each tuple has two time stamps u Time stamps conferred special status n Temporal algebra that supports temporal manipulations u Closed algebra u Complete for the temporal conditions found in protocols

Chronus Time. Line SQL (TL-SQL) GRAIN WEEK SELECT 2 ND problem_name FROM problems_table WHERE

Chronus Time. Line SQL (TL-SQL) GRAIN WEEK SELECT 2 ND problem_name FROM problems_table WHERE problem_name = ‘Hb’ WHENSTART_TIME BEFORE 1/1/99

Coupling RÉSUMÉ and Chronus n n Integrates temporal abstraction and temporal query processing Allows

Coupling RÉSUMÉ and Chronus n n Integrates temporal abstraction and temporal query processing Allows retrieval of summaries of clinical data using time-oriented conditions Modify the standard dose of AZT for a patient if, during treatment with the protocol, the patient experiences a second episode of moderate anemia that has persisted for more than two weeks

SQLA Interface Language n n n Based on SQL Supports temporal queries Detects when

SQLA Interface Language n n n Based on SQL Supports temporal queries Detects when abstractions are requested and computes them on the fly GRAIN CONTEXT SELECT FROM WHERE WHEN WEEK CCTG-522 2 ND problem_name problems_table problem_name = ‘Hb. State’ and value = ‘moderate anemia’ DURATION (start, stop) > 2

Query-Evaluation Algorithm

Query-Evaluation Algorithm

A Detailed Example GRAIN CONTEXT SELECT FROM WHERE WHEN WEEK CCTG-522 2 ND problem_name

A Detailed Example GRAIN CONTEXT SELECT FROM WHERE WHEN WEEK CCTG-522 2 ND problem_name problems_table problem_name = ‘Hb. State’ and value = ‘moderate anemia’ DURATION (start, stop) > 2

Loading the Domain Knowledge n n Examine the context clause of the SQLA statement,

Loading the Domain Knowledge n n Examine the context clause of the SQLA statement, which contains a reference to a knowledge base Use the reference to locate and load the appropriate knowledge base

Detecting the Need for Abstractions n Find non-SQLA terms in WHERE clause Blood State

Detecting the Need for Abstractions n Find non-SQLA terms in WHERE clause Blood State (“Hb. State” and “moderate anemia”) n n Look up terms in RESUME KB If look-up succeeds, Tzolkin needs to compute abstractions (“Hb. State”) Hb. State Hb WBCState WBC Plt. State Plt

Loading the Data Primitives n Locate the requested abstraction in the RESUME KB Blood

Loading the Data Primitives n Locate the requested abstraction in the RESUME KB Blood State (“Hb. State”) n Find the primitive parameters (leaves of the tree) below it (“Hb”) n Load all patient data of these parameter types into RESUME Hb. State Hb WBCState WBC Plt. State Plt

Generating the Interpretation Contexts within RÉSUMÉ n Find the types of events and abstractions

Generating the Interpretation Contexts within RÉSUMÉ n Find the types of events and abstractions that can induce a context (via a dynamic induction relation of contexts) (context CCTG 522 can be induced by event: “enroll-CCTG 522”) n Locate patient-specific instances of these events (patient enrolled in CCTG 522 on 10/10/1999) n n Compute all abstractions that can induce a context (recursive process) RESUME will then generate the appropriate contexts

Invoking RÉSUMÉ and Chronus n Execute RESUME to compute the requested abstractions u u

Invoking RÉSUMÉ and Chronus n Execute RESUME to compute the requested abstractions u u n The computed abstractions are stored in the database RESUME signals Tzolkin when it is done Then execute Chronus to retrieve the results

Future Research Directions n n Enhancement of the query language Addition of truth-maintenance capabilities

Future Research Directions n n Enhancement of the query language Addition of truth-maintenance capabilities to the database Addition of “what-if” query support Provision of complete dynamic (goal-directed) query computation