(+971) 8007 4267
(+91) 946 340 7140
(+1) 628 432 4305
Engineering Partner: Idea Usher
Multimodal Data Processing: The AI needed to analyze both unstructured data (patient-submitted wound images, operative notes) and structured data (vital signs, labs, comorbidities).
Workflow Disruption & Alert Fatigue: Surgeons rarely log into third-party dashboards. The predictive insights had to live natively within their existing EHR workflows.
Complex Interoperability: The platform required deep, bi-directional integration with both Epic and Athenahealth, necessitating a modern FHIR-based approach to bypass legacy data silos.
Phase 1: Multimodal AI & Edge Computer Vision
Two-Stage Vision Pipeline: For patients recovering at home, we developed a mobile-optimized computer vision model. When a patient uploads a wound photo, the first AI layer confirms the presence of a surgical incision, filtering out poor-quality or irrelevant images. The second layer utilizes deep learning to evaluate the incision for early signs of erythema (redness), dehiscence (wound separation), or purulent drainage.
Natural Language Processing (NLP) Engine: Because structured EHR data alone misses crucial context, we deployed a custom NLP engine to ingest unstructured clinical text. It analyzes operative reports and nursing notes to extract hidden risk factors, such as wound closure techniques or intraoperative complications, feeding this intelligence into the broader predictive model.
Phase 2: SMART on FHIR & CDS Hooks Integration
To ensure the AI system acted as an invisible assistant rather than a disruptive software tool, Idea Usher built the architecture entirely on modern interoperability standards.
Semantic Mapping: The AI’s predictive outputs are automatically converted into standardized HL7 FHIR RiskAssessment and Observation resources.
Clinical Decision Support (CDS): Instead of forcing doctors to open a separate app, we utilized CDS Hooks. When a surgeon opens a patient's chart in the EHR, a "hook" triggers our AI in the background. If the patient is at high risk for a 7-day or 30-day SSI, a non-intrusive "Response Card" appears directly within the EHR UI, detailing the infection risk percentage and the key contributing factors.
Integrating with Epic
The Challenge: Epic environments are highly customized per hospital system, and bombarding a surgeon’s In-Basket with every minor data fluctuation leads to severe alert fatigue.
The Idea Usher Solution: We integrated the platform via Epic’s Connection Hub using a SMART on FHIR app launch. To combat alert fatigue, we engineered a dynamic thresholding middleware. The system only pushes an active alert to the Epic In-Basket if the AI detects a statistically significant delta in the patient's infection risk trajectory, while routine updates are silently logged to the patient's flowsheets for review during rounds.
Integrating with Athenahealth
The Challenge: Athenahealth's strict API rate limits and webhook SLAs make continuous remote patient monitoring difficult, as ingesting high volumes of daily patient data can easily throttle the system.
The Idea Usher Solution: We deployed a secure, asynchronous message-queuing infrastructure (using Apache Kafka). Instead of continuously pinging Athenahealth, the middleware batches routine patient updates (like daily temperature logs or low-risk wound images) and syncs them during off-peak hours. However, if the AI detects an acute risk—such as a sudden spike in wound inflammation—it utilizes priority event notifications to bypass the queue and instantly update the provider.
Improved Patient Outcomes: The AI pipeline accurately identified high-risk SSI trajectories up to 80% earlier than standard manual chart reviews, drastically reducing 30-day hospital readmissions.
Maximized Clinical Efficiency: By automating SSI surveillance and integrating directly via CDS Hooks, infection control teams saved hours previously spent on manual data abstraction.
Seamless Scalability: The adherence to FHIR and SMART standards meant the platform could be rapidly deployed across various hospital networks, ensuring full HIPAA compliance and zero disruption to established clinical workflows.
* Terms & conditions apply
- F 337, Industrial Focal Point, Phase 8B, Industrial Area, Sector 74, SAS Nagar, Punjab 160055
- SCF 98, Phase 11, Sector-67 Mohali, 160062
- [email protected]
- (+1) 628 432 4305
