was successfully added to your cart.

Big Data HealthcareNatural Language Processing

Lessons Learned From Five Real-World Case Studies 

By May 25, 2021May 27th, 2021No Comments

In our previous post, Key Takeaways from the Healthcare NLP Summit, we spoke about the Healthcare NLP Summit, a two-day event that brought thought leaders from business and academia to share their opinions on the emergence of AI in the healthcare industry, allowing medical professionals to better aid patients and accelerate service delivery.

The virtual event captured discussions and learnings on diverse topics indispensable to the evolving healthcare industry. This continuation post compiles a list of the most enriching and key takeaways from the two-day event. Continue reading for a rundown on five of this year’s Healthcare NLP Summit’s most impacting sessions:

 

 

What We Learned From Five Real-World Case Studies

The 2021 Healthcare NLP Summit speakers shared a wealth of knowledge, intelligent insights, and use cases. Listed below are the five sessions that were particularly relevant:

 

Connecting the Dots in Clinical Document Understanding and Information Extraction

Day two of the Healthcare NLP Summit began with a keynote by Veysel Kocaman, Principal Data Scientist and ML Engineer at John Snow Labs. He shared his insights on how EHR data extraction is imperative to clinical Named Entity Recognition (NER) for clinical data linking, assertion status, de-identification, and relation extraction. Veysel brought forward how recent advances in AI, ML, and NLP are helping healthcare professionals to make decisions based on the data for an accurate screening diagnosis and treatment. 

Veysel also shared a comprehensive competitor analysis on Spark NLP Clinical Models and how they fared with AWS Medical Comprehend and GCP Healthcare API.

Watch this session for free here.

 

Best Practices in Improving NLP Accuracy For Clinical Use Cases

Rajesh Chamarthi, a Senior Big Data Engineer at MetiStream, and Veysel Kocaman, Principal Data Scientist and ML Engineer at John Snow Labs, led this session.

Rajesh began by sharing a high-level overview of MetiStream’s Ember platform and described a plethora of use cases on the providers, bio and life sciences, and payers’ side. He explained bio and life sciences could use the Ember NLP product to accelerate drug development and to be able to identify better patient information for clinical trials. Another interesting clinical use case he shared was on the Adenoma Detection Rate (ADR) quality measure. Rajesh and Veysel explained the greedy mode in the NER converter using high-grade dysplasia of the colon as a single entity. 

Rajesh discussed the identification of correct SNOMED concepts for NER. He was joined by Veysel on how John Snow Labs collaborated to design and implement models, as well as apply best practices and processes to improve clinical entity and code accuracy to bolster significant medical and quality use cases in healthcare.

Watch this session for free here.

 

Beyond Context: Answering Deeper Questions By Combining Spark NLP and Graph Database Analytics

The next session featured Abhishek Mehta, Director of Field Engineering at TigerGraph, and Christian Kasim Loan, Senior Data Scientist at John Snow Labs. In this informative session, Abhishek shared market projections by Reuters on the BI industry and added that Reuters estimates the global revenue of BI would rise to $29.48 billion by 2022. 

About 80% of business data is unstructured across the business domains. Healthcare records contain data on biosciences comorbidity, insurance records, lab informatics information, sales records, physicians’ information, and health center admissions, all of which are interrelated datasets on the Healthcare Network graph. Abhishek continued the session by explaining how healthcare information is siloed, supported by diverse data formats and data privacy concerns. 

While John Snow Labs’ solutions extract information and knowledge from multiple EHR datasets, TigerGraph is leveraged to bring in connectivity that addresses the speed, the scale of information for pattern recognition, and to perform deep text analytics at scale. Christian demonstrated how Spark NLP can aid in cleaning raw data (5.4 million tweets and 2 million tweets with user profiles) collected from social media for the Tiger Analytics solution to be mined for intelligent insights.

Watch this session for free here.

 

Accelerating Clinical Risk Adjustment Through Natural Language Processing

The fourth session featured Fola Soyoye, Director of Product Management at iQuartic, and Alina Petukhova, Data Scientist at John Snow Labs. This session covered:

  • Risk adjustment overview
  • Why is risk adjustment an NLP problem
  • Case study: Computer-Assisted Coding with a Solution Approach

Healthcare organizations are increasingly using natural language processing (NLP) technology to improve their risk management and value-based care programs. Over the years, the number of healthcare systems that have embraced NLP technology has expanded substantially, resulting in an increased risk capture veracity.

This session focused on the issues of risk adjustment. It showed how Spark NLP for Healthcare offers the potential to speed up the evaluation and extraction of diagnosis codes from unstructured medical records.

Watch this session for free here.

 

Explainable Deep Learning For Automated Extraction From Free-Text Medical Reports to Enable Precision Healthcare

Our fifth feature was presented by Vishakha Sharma, Principal Data Scientist at Roche, one of John Snow Labs’ customers. Vishakha explained the NAVIFY Tumor Board—a cloud-based workflow product that integrates and displays aggregated data to a unified patient dashboard for oncology caregivers to provide optimal treatment. 

She continued by sharing the challenges of unstructured reports’ data extraction, ranging from jargon to handwritten notes that can be very time-consuming, expensive, and error-prone for manual curation and how NLP and OCR support NAVIFY to address these challenges. 

Concluding the session, Vishakha summarized how Roche applies Spark NLP to extract clinical intelligence from pathology, radiology, and genomics reports.

Watch this session for free here.

 

Looking Forward

The above lessons showcase a few of the 35 sessions that took place at the Healthcare NLP Summit. Please visit the NLP Summit Website for on-demand access to all sessions.

Want to join in on a live NLP Summit event? Don’t miss the date for the next virtual conference from October 5-7, 2021. To secure your spot and receive regular updates on the next editions of this event series, register today.