Gather clinical insights from the MIMIC-IV dataset with CogStack
Preview the pipeline that is deployed using the MIMIC-IV dataset to hosting clinical insights and visualisations on CogStack Dashboards
The entire process is listed via: ingestion, hosting, transforming, annotating, and finally visualising the MIMIC-IV dataset
Discover the MedCAT demo trained on the MIMIC-IV dataset to unlock real-world clinical insights
Enter free text clinincal notes and generate annotations that can be explored within the webpage
Annotations follow the SNOMED CT ontology, and can be used to identify clincical concepts such as disorders, findings, procedures, and medications
CogStack Dashboards hosts MIMIC-IV, via single sign on with PhysioNet.
Discover the data of the MIMIC-IV dataset, with all data accross all tables being freely accessible to credentialed users.
Explore additional indexes, annotations that been generated via CogStack MedCAT.
CogStack Dashboards hosts MIMIC-IV, via single sign on with PhysioNet.
Discover the data of the MIMIC-IV dataset, with all data accross all tables being freely accessible to credentialed users.
Explore additional indexes, with annotations that been generated via CogStack MedCAT.
curl -X GET "https://cogstackdashboards.sites.er.kcl.ac.uk/auth/api/proxy/discharge_annotations/_search" \
-H "Content-Type: application/json" \
-H "Username: <PhysioNet username>" \
-d '{
"size": 1,
"query": {
"match_all": {}
}
}'Sample Response
{ ...
"nlp.start" : 12205,
"nlp.end" : 12208,
"nlp.cui" : "277889008",
"nlp.acc" : 0.8735785540901717,
"Presence" : {
"confidence" : 0.9999730587005615,
"name" : "Presence",
"value" : "True"
}
... }CogStack Cohort is a web application that lets users to quickly identify patient cohorts by writing a natural lanugage definition
Converts natural language queries into structured JSON (including phenotypes, demographics and temporality)
Execute generated queries to match the cohort from health record annotations
Querying can be done via multiple filters; such as age, gender, and ethnicity to more complex filters such as clincal terms.
We provide a demo of Cogstack Cohort, that presents the cohort information of the patients present in the MIMIC-IV dataset.
Foresight is a generative AI that combines converts EHRs into structured and coded longitudal predictions of long term pateint outcomes.
Foresight is a novel transformer-based pipeline that uses both entity recognition tools (MedCAT), and provides probabalistic forecasts for future medical events such as disorders, findings, and interventions.
A demo of foresight is publicly available, which is trained on the MIMIC dataset. Allowing users to explore synthetic patient trajectories, disease progression, and future event forecasting in an interactive setting.
Glimpse into the future of digital health twins, virtual trials, and scalable longitudal health.
Seamlessly search across unstructured clinical text (notes and radiology reports) and structured data such as laboratory results through a single unified interface.
The agent dynamically builds multi-step retrieval workflows that filter patients via structured data first, then surfacing relevant clinical narratives for deeper analysis.
