Our Technology

Artificial Intelligence. Naturally.

Our founders spent years at IBM Watson, and it shows.

Simply put, AI is in our DNA. NeuralFrame’s approach to medical AI is based on our founders’ deep expertise in building medical language models and extending existing medical ontologies.

The UMLS (Unified Medical Language System) is a great place to start, with more than 130 source vocabularies and 5-level taxonomy of circa 150 entity types. But its internal inconsistencies and imperfect relationship rules require a more informed and rigorous approach. This is exactly what NeuralFrame brings to the table. Our expertise in labeling unstructured medical text has allowed us to build a unique Knowledge Iteration Engine and domain-specific CORE Models that go beyond traditional approaches to medical NLP (Natural Language Processing).

And the more data we gather (see the section on FHIR below), the more accurate and powerful our medical AI becomes, as we continually improve our framework with both explicit and implicit feedback. Our expert-weighted knowledge graphs, combined with state-of-the-art language models, allow rapid learning and adaptation, putting NeuralFrame squarely at the forefront of medical AI innovation.

This may all sound great, but what does it mean in the real world? Well, here’s a for-instance: When we analyze usage statistics from our industry-leading cancer registry product, KACI, we find the following time savings for ODS’s in abstracting cases from the EMR:

KACI’s fast and accurate automatic abstraction frees up ODS’s to focus on complex analysis and reporting instead of hunting through synoptics reports and other structured and unstructured medical documents.

This observation about KACI brings us to a final thought on NeuralFrame’s approach to medical data. Founded in 2017, we brought together an expert team of clinicians, systems engineers, and AI specialists to build better products for the medical community, with AI at the core. KACI is not a software application with an AI engine retrofitted as a bolt-on. KACI, and every product that comes from NeuralFrame’s software lab, has been engineered from the ground up around a deep understanding of the integration of AI with medical data. By building our products with AI as the core foundation rather than as an afterthought, we are able to deliver a seamless and uniquely valuable experience for healthcare professionals.

Drinking from the FHIR Hose.

Next-generation architecture. Massive record counts.

The Fast Healthcare Interoperability Resources (FHIR) standard was developed by the Health Level Seven International (HL7) organization, a not-for-profit standards development organization focused on healthcare IT interoperability. The genesis of FHIR can be traced back to the mid-2000s when the need for a more modern and flexible healthcare data standard became evident.

The benefits of using FHIR to exchange data with EMR systems are manifold. FHIR’s standardized data models provide a consistent and structured way to represent clinical information. This consistency enhances data quality and accuracy. Furthermore, FHIR is fast. Its RESTful APIs allow for efficient data retrieval, searching, and updating, leading to timely reporting. And let’s face it, the sooner you can report medical data, the more likely your data will be useful to the designers of clinical trials, and the patients they serve. In this business, saving time means saving lives.

All NeuralFrame products are FHIR-native, meaning that our systems architecture has been designed from the get-go with FHIR interoperability in mind. The result is that we can use FHIR to connect with all relevant EMR resources, from your primary EMR to specialty EMR and other diagnostic sources for radiation, surgery, systemic therapy, and pathology. Every week, NeuralFrame automatically ingests millions of EMR records for its customers (see our landing page for the latest counts for this week). These records are then passed through our proprietary medical AI models (see section on AI above) to identify Casefinding and Follow-up data. Combined with KACI, our cancer registry software product, the benefits are immediately apparent:

Casefinding

ODS Workflow Improvement

Follow-up

Registry Management

Casefinding – KACI’s EMR integration directly accesses pathology reports, eliminating the need for separate pathology surveillance software. KACI combines pathology with other EMR resources from the clinic and billing to deliver the most accurate records possible, removing duplicates and not-reportables, without missing cases. KACI automation typically reduces Casefinding time by more than 80%.

ODS Workflow Improvement – After searching patient records, the Casefinding process captures and pre-populates circa 40% of the data fields, reducing manual input time and eliminating transcription errors. Since gains in this area impact the entire ODS team, even a conservative estimated efficiency gain of 12% delivers significant economic benefit.

Follow-up – Follow-up processes that require manual effort are expensive and prone to backlog. KACI integrates multiple EMR resources, including external resources such as Epic’s Care Everywhere, to update Date-of-Last-Contact. The integrated Social Security Death Index (Death Master File – DMF) provides timely updates for expired patients. According to the NCRA Workforce Study, Follow-up at the average hospital requires 2.5 FTE per 6,000 new diagnoses. KACI reduces Follow-up labor by 80-90%.

Registry Management – Hospital workflows that employ manual processes cost significant registry management time. KACI’s EMR integration reduces QA issues, saving management time. In addition, KACI’s QA dashboards allow ODS’s to self-check. KACI integrates case assignment and abstract notes to eliminate spreadsheets. Users can save up to 90% of case review time.

Partnership for the Future.

The ever-expanding cloud.

NeuralFrame is currently building custom extensions of its KACI Platform for some of the largest healthcare organizations in North America. These forward-looking institutions have recognized NeuralFrame as a strategic partner in their efforts to future-proof their registry requirements, while delivering a scalable and interoperable cloud-based data backbone that extends well beyond current registry operations.

Our expertise in applying advanced medical language models and leveraging the power of FHIR enables these organizations to unlock the full potential of their data. Here are some of the ways we are helping our customers to adapt and thrive in the rapidly evolving healthcare landscape:

Billing

Patient Search & Matching

Patient Outcomes

Data Warehousing

Billing – Enabling Administrators to generate accurate and detailed patient billing.

Patient Search & Matching – Enabling Life-Science organizations to identify patients as candidates for research studies and patient cohorts in real time, followed by clinical trial support and data presentation showing fairness of clinical trial programs.

Patient Outcomes – Enabling Life-Science organizations to track research study participants and monitor patient outcomes as they respond to new treatments and therapies.

Data Warehousing – Enabling academic centers to provide timely, structured, and rich datasets to academic analytics departments, exporting NeuralFrame data on a nightly basis for interrogation by scheduled and adhoc SQL queries.