Analytics Drawer
- Emory Fry
The Clinical Analytics Drawer communicates a patient's relative risk for developing one or more diseases. A high relative risk score with a narrow confidence interval, however, should not be misinterpreted as diagnostic. In other words, a patient with a 90% probability of developing posttraumatic stress disorder, may fortunately not actually have the disease. To facilitate an appropriate diagnostic workup for the condition in question, a provider will be able to access the diagnostic guide functionality provided by the unified collaboration portal. This clinical decision support tool, launched by clicking the quote starts diagnostic guide” button available in the drawer, initiates a predictive model tool that facilitates diagnostic decision-making.
When launched, the diagnostic guide replaces the center canvas area of the portal with a graphical depiction of the steps and choices available to a provider while making decisions regarding a diagnostic workup. In the posttraumatic stress disorder use case, the tool initially displays the current probability of disease, the calculated confidence interval, and a select list of diagnostic tests and interventions that might be appropriate during the workup. Options might include performing its psychometric assessment, obtaining relevant diagnostic imaging studies, completing survey questionnaires, or ordering laboratory tests. For each of these, the diagnostic guide calculates a utility score that reflects the models prediction that ordering the test will contribute to making the diagnosis. A utility score is simply a reflection of the benefit and intervention may have for achieving a particular clinical goal. Diagnostic utility, the benefit afforded to making a diagnosis, is the most prevalent and relevant utility score. However, for a particular patient, cost-effectiveness or patient comfort might be the principal goal. In such situations, the predictive model used for a particular disease, will need to calculate financial or pain utility respectively. Only diagnostic utility is currently in scope for this project, but the requirement for calculating and displaying alternative utility scores is clearly evident.
The diagnostic guide displays explicitly the top 3 or 5 choices to the provider. Other tests or interventions that might be appropriate for a given diagnostic workup are available to the provider via the quote low utility choices” interface control. Additionally, in the entire repertoire of the EMR ordering capability will be made available to the provider who selects the quote other choices” control even if these interventions were not recognized by the predictive model as being relevant to the diagnostic workup.
When a provider selects one of the options presented, the system will display summary data describing the intervention, its potential relevance for the workup, and whether the choice is available at the local facility and/or the local community. After reviewing this information, the provider may accept the choice and trigger the subsequent workflow. For example, if the option is a lab oratory test, the appropriate order will be placed in the system and electronically activated on the provider's behalf. Alternatively, if the choice is to complete a survey questionnaire, the survey form will be immediately displayed for the provider to complete. Finally, if the option recommends a particular physical exam, the system will allow the clinician to record the result after the human task has been completed.
When the provider has finished selecting the options they feel are appropriate, the diagnostic guide will display the current status of each choice and whether the results are available are pending. As tasks cannot be expected to be completed at the same time or even day, the system will retain the current diagnostic guide state at all times, allowing a provider who leaves the tool for any reason to resume where they left off when they return at a later date.
At some point all chosen interventions will have been resulted, and the diagnostic guide will allow the provider to advance to the next iteration. Using the newly resulted information, the guide will recalculate the patient's diagnostic certainty score and display the next 3 to 5 choices that are appropriate. It will also reanalyze the diagnostic utility of those options given the results of the previous iteration.
After one or more iterative cycles, the patient's diagnostic certainty score will be 100%, or high enough that further workup has no clinical benefit, which is a common decision that must be made with diagnoses of exclusion. At this point, the provider may end the diagnostic guide session, and select from a tailored list, and record and appropriately coded diagnosis. The provider may, alternatively, cancel a guide session at any time without recording a new condition.
Diagnostic guide sessions are patient based and not unique to a particular provider. This means that when a patient record is accessed by one or more providers, each of them can see and contribute to any active disease guide session. A guide workflow, therefore, is a powerful collaboration and coordination tool for aligning the diagnostic efforts of the clinical team, communicating work up strategy, and reducing unnecessary duplicative efforts.
Unlike traditional clinical pathway tools or structured prescriptive diagnostic aids, the Sirota diagnostic guide focuses on providing the qualitative information, expressed as utility scores, that a provider needs to evaluate the appropriateness of a choice for the attainment of a particular goal. Reflecting the common workflow found in clinical medicine, the guide supports a semistructured, iterative cognitive workflow with frequent reassessments. It is a goal oriented approach, that encourages evidence-based decision-making while retaining provider flexibility and autonomy for delivering individualized patient care.
The diagnostic guide provides an additional benefit for clinical quality improvement initiatives in that it records through every iteration the patient's disease risk, the choices made available to the provider, the utility scores at that stage of the workup, and the choices that were eventually made. At any time during the diagnostic workup, the provider can review their decision making process and critically review their choices. Alternatively, institutional process improvement teams can aggregate diagnostic guide logs to evaluate provider performance, workflow impediments, or institutional barriers to quality care. The diagnostic guide offers a powerful tool for collecting information relevant to clinical operations into meaningful than quality improvement analysis.