This version of CTcue is an update from version 4.3.0 and contains a number of new functionalities, improvements and various bug fixes. Below you will first find a brief overview. Below that, you can read the extended release notes with more detail about some of the main changes in this version.
Features
- In patient finder, you can now search for patients that match a minimum X out of Y terms in a group.
Improvements
- Auto-complete for nested categories (admission period, form entries, and surgery procedures) now include the parent filters to give better suggestions
- For connector exports: The answer id is always exported, even if no results are found. The section id and name are exported as well
- The preview for report plus terms excludes unprocessed reports
- Report plus results are now sorted by their relevance
- Highlights in report plus texts are now colored to reflect the relevance of the corresponding concepts. This makes it easier to link the concepts in the confidence table to the matching concept in the text snippet and interpret the context.
- The English (UK), Czech, Spanish, Italian, German and French translations have been updated
Bug fixes
- Data validation now correctly synchronizes if the questions at the last synchronization have been completely replaced
- Users are denied access when using organization login if there is not at least one applicable role mapping
- The reports plus category icon in the relative date filter term selection list is no longer cut off
- The validation label column in exported CSV and Excel files is no longer duplicated
Extended release notes
Below is a more detailed overview of some of the bigger or impactful changes in this version.
Minimum match option for groups in the Patient Finder
Patients may use a variety of medications to treat one or more diagnoses at the same time. Patients who use a combination of different medications could be of special interest when building a patient cohort. In some of these cases you may not be looking for specific combinations, but for patients who use for example at least three out of a group of five medications.
To allow users to find these patients, we have introduced the option to determine a minimum match option for groups in the Patient Finder. When creating a group with multiple terms, you can choose whether patients must match all terms in the group, or whether they should match a minimum number of terms out of the group. This can be used both for the included and the excluded patients.
Note that nested groups (group within a group) will be treated as one term when matching a minimum number of terms.
This is how to use the feature:
- Create a group for the included or excluded criteria in the Patient Finder
- Add the terms for the group
- Change the dropdown field at the top of the group from must have "all" to must have “any”
- New fields will appear where you can define how many terms out of the total group must match
Auto-complete suggestions for nested categories
When searching for terms in CTcue, we provide auto-complete suggestions based on the most used terms, which makes it fast and easy to build queries. When using nested terms (admission periods, form responses, or surgery procedures), we now filter the auto-complete suggestions based on the terms used in the parent term. This will help make the suggestions more applicable to what you looking for.
Export unknown answers with an AnswerID
When you export the data for a multiple choice question in the Data Collector using a MSSQL connector, we provide a unique ID for each answer option. There is also a fallback group for multiple choice questions when there are no results for the defined answer options, but since this field represents the lack of data, there was no ID linked to it. The lack of data, however, is also valuable data to capture, so therefore we have now added an AnswerID to this field as well.
Reports plus updates
Reports plus is still an experimental feature in the Patient Finder that is being tested with a few hospitals before rolling out to the Data Collector and opening up to the broader public. Based on the initial feedback we have included a number of improvements in release 4.3.1.
In the preview for reports plus terms we no longer show unprocessed reports, because they are not helpful in refining the search query and may cause confusion with users. Unprocessed reports are still shown when performing a full search.
Reports plus terms are now sorted by their relevance first instead of by their confidence. Although the confidence could be very high for some concepts, we may still mark their relevance as unsure or irrelevant, because of the context (e.g. pertaining to a family member or falling outside of a date range). To help focus on the results most relevant to a user, we now sort the relevant concepts first, then unsure concepts, and irrelevant concepts last. Concepts with the same relevance are still sorted by their confidences. In case of a tie between both relevance and confidence we sort by extracted start date if available or report date (newest to oldest). In the single patient view, the option is added to change the sorting by highest or lowest relevance.
Despite having a high confidence score, we may classify the relevance as unsure or irrelevant based on the context of the term. For example, we classify reports plus terms as irrelevant when it pertains to a family member or when it is negated. And we classify terms as unsure if it occurs in the future. To help better understand why reports plus terms are classified differently from what would be expected based on the confidence buckets, we now display these context labels in the results table on the single patient page.
To make it easier to understand the reports plus terms in their context, the highlighting of the terms now matches the relevance category. That means that relevant terms are highlighted green, unsure terms are highlighted orange and irrelevant terms are highlighted red in the report text snippets. Also, when hovering over a reports plus term, the relevance, confidence score and context label are displayed.