User:Gjsikora/sandbox

Lexenco Clinical Language Encoder (CLE)™ is a clinical Natural Language Processing (NLP) that harvests clinical information and medical codes (including Current Procedural Terminology (CPT)™, Hierarchical Condition Category (HCC), Healthcare Common Procedure Coding System (HCPCS), ICD-10, RxNorm and SNOMED CT) from unstructured text. Deployed as a Software as a Service (SaaS) software deployment model, integration requirements into variant healthcare IT solutions are minimal. healthcare IT solution providers utilize clinical Natural Language Processing (NLP) to increase documentation quality and understanding, optimize claims reimbursements, and improved productivity.

Background
Within healthcare IT solutions, vast amounts of clinical information is represented in unstructured text form as a product of narrative-based system interaction, and textual-based clinical reports and publications. The challenge for Clinical Documentation Improvement (CDI) professionals is to ensure their assigned codes derived from these unstructured resources accurately reflects the patients’ clinical status and provided care for quality reporting and assuring that the organization captures all of their entitled reimbursement charges, meanwhile optimizing coding productivity. In answering this need, a common goal of healthcare IT solutions providers is to integrate Computer Assisted Coding (CAC)  medical autocoding technology within their solution to automatically detect and extract clinical information and medical codes from these unstructured resources. In turn, their customer benefits are streamlined revenue-cycle processes while becoming more compliant with the increasingly complex payer and quality reporting requirements, bolstering their product amongst competitors.

History
Lexenco CLE™ development originated at Progeny Systems Corp. under an OSD06-H09, "Natural Language Processing" Small Business Innovation Research (SBIR) project, initially funded in 2006. The objective was to design and develop an NLP engine that links Department of Defense (DOD)'s Armed Forced Health Longitudinal Technology Application(AHLTA) Electronic Health Record (EHR) medical records to Unified Medical Language System (UMLS) medical code ontologies. The goal was to process unstructured text note sections of medical records stored within the AHLTA EHR Clinical Data Repository (CDR) to capture both contextual and structured terms in support of surveillance and data mining use-cases.

Through 2013, several commercial clinical NLPs, including clinical Text Analysis and Knowledge Extraction System (cTAKES) – an open-source Apache solution, were evaluated with respect to satisfying AHLTA use-case requirements. These clinical NLPs performed satisfactorily with atomic medical codes such as ICD-9 and SNOMED CT) where codes represent single concepts. During the following 2013 to 2015 period, the healthcare industry matured ICD-10 in which AHLTA had to support. Unlike ICD-9, the ICD-10 medical codes are complex with intertwined laterality, specificity, negation, temporality and demographics across sub-concepts. In 2013, these ICD-10 complex requirements drove the genesis of the Lexenco CLE™ product known today, built on top of open-source components, comprising Unstructured Information Management Architecture (UIMA), OpenNLP and UMLS.

Meanwhile, the future of the legacy AHLTA system was uncertain. For various cost and performance considerations, replacement strategies were underway. Early in 2011, the Secretaries of DOD and Veterans Affairs (VA) agreed to work cooperatively on developing one common integrated Electronic Health Record (iEHR) system by 2017. An Interagency Program Office (IPO) was formed by the National Defense Authorization act assumed responsibilities for the iEHR. Unfortunately, disputes erupted between VA and DOD because each had its own domain and ways of doing things, and neither was willing to cede ownership. Early in 2103, the Secretaries of both Departments announced that instead of building a single integrated EHR, both DOD and VA will concentrate on integrating VA and DOD health data by focusing on interoperability and using existing technological solutions. However, DOD health professionals continued to find AHLTA to be difficult to use, slow, and frequently subject to crashing. As such, in 2013 DOD began taking bids to overhaul the system. In 2015, DOD awarded the contract to the Leidos Partnership for Defense Health, a consortium of EHR manager Cerner; management consulting and professional services company Accenture Federal Services; and engineering and technical government contractor Leidos.

In response to the changing landscape of the DOD EHR from a government owned AHLTA system to a commercial Cerner system, the Lexenco CLE™ architecture changed from a one-off AHLTA integration to a Software as a Service (SaaS) software deployment model, enabling easy integration into variant healthcare IT solution opportunities across the commercial marketplace.

Capability

 * Extracted Clinical Code Sets: CPT™, HCPCS, ICD-9, ICD-10, (Rx)HCC, RxNorm and SNOMED CT.


 * Section Inclusion/Exclusion: Select the medical record section types to process, e.g. include “History of Present Illness” and exclude “Past Medical History”.


 * Clinical Semantic Compression: Clinical terms within source text are semantically compressed to candidate clinical concepts, such as “amobarbital”, “Amsal”, and hundreds of other terms for “barbiturates” in ICD-10-CM T42.3X1 “Poisoning by barbiturates, accidental”.


 * Section Aware: The section of the clinical concept term is used to improve clinical code extraction, such as ICD-10-CM Z80.6 “Family history of leukemia” vice C91 through C95 codes.


 * Negation Aware: Negation expression detection with corresponding clinical concept term tagged, used to negate output and extract clinical codes with negation, such as ICD-10-CM A37.90 “Whooping cough, unspecified species without pneumonia”.


 * Source Text Input Formats: Text field, and text, Word and PDF documents.


 * Context Aware: Experiencer and temporality aware processing provides context understanding of the extracted clinical concepts.


 * Code Concept Decomposition: Category type code sets such as ICD-10-CM have compounded definitions that are broken down to atomic clinical concepts then matched separately against a respective ontology, and composed across a section, such as S72.001B “Fracture of unspecified part of neck of right femur, initial encounter for open fracture type I or II” – “minimal/ moderate soft tissue injury”, etc.


 * Returned Information: Options for returned information of the extracted code includes relevance, category, source location and section, coding advice, and clinical concept.


 * Clinical Code Navigation: Navigate across clinical code set hierarchy by clinical code, get code by clinical concept, and get tree for a set of clinical codes.