top of page

Propel Health AI - Empowering the Healthcare Ecosystem with Data and AI

Our Services

Data Ingestion and Storage

Propel integrates with medical record systems and data sources to securely collect clinical patient records found in both structured and unstructured formats as well as where applicable the associated multi-modal datasets , such as pathology results, medical images (DICOM) and genomics images. Data is encrypted in-transit and persisted within customer-specific isolated storage, protected by restrictive read permissions and customer-specific encryption keys. 

Data Harmonization & Interoperability 

Patient records are harmonised (processed, de-duplicated, and reconciled) into a central patient record store in FHIR format, with corresponding consent records. To facilitate data harmonisation, Propel's harmonization and interoperability pipeline defines mappings from each source-aligned dataset to the Propel FHIR store, defines patient record matching rules for each dataset based on MRN/UR numbers and harmonizes patient records with consent records.

collaboration-white.png

Data Deidentification

Propel facilitates de-identification of all unstructured and structured clinical records as well as all radiological images stored in DICOM format. De-identification is performed according to the US HIPAA Safe Harbour method in Section 164.514(b)(2) of the US HIPAA Privacy Rule, which specifies 18 identifiers to be removed, or via specific de-identification configurations. 

document-white.png

Unstructured to Structured Data Transformation

Our cutting edge data ingestion capability uniquely employs Large Language Model (LLM) capabilities to effectively structure unstructured data such as clinical notes and medical records. Our infrastructure allows for extracting key patient details such as diagnoses, treatments, symptom, medications and dosage from unstructured sources and using that to enrich structured FHIR records that can be queried and analysed leveraging our cohort builder and analytics workbench capabilities

Cohort Builder

Propel allows the assembly of de-identified data into cohorts based on key patient characteristics, such as diagnosis, medication, family history, procedures, and more, making them available for collaborative analysis. A cohort, a curated collection of patient records and related data, aligns with specific consent documentation, ethical approval and eligibility criteria for therapeutic or clinical trial participation.  Access to catalogue search and cohort creation is granted exclusively to authorized users, ensuring data security and privacy.

Analytics Workbench

Our analytics workbenches are a secure data enclave, facilitating collaboration by providing a space where authorized parties can work with specifically consented data. The workbench allows collaborators to interact with cohort datasets through a secure user interface. The workbench provides interactive notebooks (Python & R), data visualisation and business intelligence tools, machine learning capabilities, workflow management and DICOM medical viewers, and  to authenticated and authorised collaborators. Fine-grained security guardrails prevent uncontrolled data egress of patient records and associated multi-modal datasets. Our machine learning capabilities also provides the infrastructure for model development, validation, deployment and monitoring with pre-built pipelines for bias checks, performance drifts and model cards.

Audit Logs

Audit interface that provides users with a comprehensive record and visibility into the activities and events occurring within their platform tenancy

security-white.png

Security and Privacy

Security and privacy are at the forefront of our mission, and incorporated into our products from the ground up, starting from the design phase all the way to testing and production release. Customer data is kept in a quarantine zone without human access, and data ingress and egress are only done via operational pipelines.
We go above and beyond the regulatory requirements for Protected Health Information (PHI) and Personally Identifiable Information (PII) for every country and region, and work together closely with our clients to meet their needs to achieve compliance. We also use best-in-class systems and tools to augment our privacy and security posture, from the in-built security architecture of Google Cloud Platform, to the cutting-edge information security detection and prevention capabilities of Wiz. Propel is working towards information security certification for ISO27001:2022 and SOC 2, and documentation are available upon request.

Seamless Integration with Healthcare Providers' Ecosystem

We understand that our clients operate within complex healthcare data ecosystems. That's why our platform is designed to seamlessly integrate with existing healthcare systems, ensuring that our clients can easily access and share data across their organization.

bottom of page