Ergin Soysal, CTO
Healthcare, as we know, is an information-dense industry but unfortunately, most of the patient information is embedded in free narratives and lacks the technology to interpret it. This often leads to an incomplete diagnosis of patients due to lack of proper information. Ergin Soysal witnessed this gap in technology and set about to bring a change through intensive research. Now, envision a healthcare solution that can emulate a human’s problem-solving capacity as well as organize data from various channels into a unified unit. This seemed like a sci-fi fantasy five years ago, but after a decade of research, Melax Technologies was established to enable just that. The company’s avant-garde natural language processing (NLP) and machine learning solution harness the power of data to maximize productivity across healthcare systems, clinical research centers, and data analytics firms. “We give end-customers a reliable, consistent, high-performance, customizable, and user-friendly NLP engine to interpret textual data,” says Soysal, CTO of Melax Technologies.
What started as a collaboration between Soysal’s team and multiple academic medical centers in Texas Medical Center is now revolutionizing the healthcare system with a customized NLP models that use advanced coding to extract useful data from treatment tests, medication, and temporal information. Melax Technologies runs its NLP solution in an integrated development environment (IDE) to reduce the time in creating, testing, comparing, and analyzing the shortcomings in algorithms. But accomplishing this feat was not easy.
We give end-customers a reliable, consistent, high-performance, customizable, and user-friendly NLP engine to interpret textual data
“After intense research, we overcame problems such as incapability to run on multi-user production environments, lack of patient data security, and the variability in data processing models,” explains Soysal.
Clinical Language Annotation, Modeling, and Processing Toolkit (CLAMP) improve the decision-making capability of healthcare providers with insightful data gathered from clinical free text. CLAMP is also the tool of choice for data analytic companies developing their own analytic solutions. CLAMP consists of three major components. CLAMP-CMD provides a robust, high-performance NLP infrastructure for clinical text processing.
CLAMP GUI provides pre-trained machine learning models, clinical dictionaries as well as syntax rules to develop custom NLP pipelines, visually. This is the actual strength of CLAMP; it enables organizations to build a state-of-the-art machine learning models based on their very own data. Since a model developed on a data from one organization may not be repeating the same performance in another organization, this allows to fine tune algorithms for a specific task, precisely. Additionally, Melax Technologies has developed an enterprise-level data monitoring tool to process data across clinical repository, data warehouse, and operational workflow on distributed NLP nodes in a given schedule. Today, the success behind Melax Technologies can be attributed to its ability to provide a reliable NLP solution that interprets biomedical data and solves the problems associated with linguistics such as textual data and language interpretation.
Soysal shares a befitting client scenario, where a medical center in Texas was re-evaluating their billing system. Unfortunately, the client found that a lot of bills pertaining to discharged patients were waiting to be processed manually and sent to the insurance company for reimbursement. Melax Technologies collaborated with the client’s developers to create an EHR integrated NLP pipeline that quickly processes each note on patients discharged, classifies patients, and provides coding suggestions to highlight certain data. At the end of a month, the client reaped the benefit of a 40 percent improvement in its overall processing time and gained substantial economic returns.
Striding head-on in the NLP race, Melax Technologies is now invested in understanding the customers’ requirements as per their data variability, research into the importance of visual analytics tools, and how it would help end-users to make sense of the extracted data.