USE CASES

CLAMP for the Pharmaceutical Industry
Discovery of biomedical knowledge from a massive scientific literature is an arduous task. By leveraging the award-winning biomedical NLP technology engine developed at Melax, scientists and researchers now can be relieved of expensive and time-consuming literature search/review, and truly focus on their innovation in research and development. For example, we are offering a systematic review tool that implements an online learning algorithm to automatically learn and classify relevant document during the review process.
Personalized cancer therapy, which provides tailored treatments based on a patient’s specific characteristics (e.g. genetic status), has shown great promise for improving outcomes for cancer patients. Meanwhile, hundreds of clinical trials are investigating drugs that target specific genetic alterations in tumors. A CLAMP based system was developed by Melax for extracting genetic alteration information for personalized cancer therapy from ClinicalTrials.gov.
CLAMP for Hospitals, Clinics, and Insurance
Melax has worked with multiple clients to develop CLAMP-based CAC systems for both ICD-10 and Hierarchical Condition Category (HCC) Codes to improve reimbursement rate and reduce the manual coding workload. A CAC system typically consists of NLP technology components such as named entity recognition, relation extraction, concept normalization and utilizes both machine learning and rule-based technologies. CAC based on the most recent ICD-10 and HCC codes improves reimbursement rate, and decreases claim denials and appeals.
Quality measures help quantify healthcare processes, outcomes, and organizational structure and systems. Melax has developed NLP-based approaches for quality measurement reporting. For example, we developed a system for extracting Venous thromboembolism (VTE) mentions from radiology reports and encoding them to SNOMED concepts, for one of the meaningful use criteria. Leveraging the award-winning algorithms implemented in CLAMP, an extremely high sensitivity of 0.98 and a reasonable PPV of 90% for the detection of VTE cases was achieved at a client site.