Desiderata for delivering NLP to accelerate healthcare AI advancement and a Mayo Clinic NLP-as-a-service implementation.

Data is foundational to high-quality artificial intelligence (AI). Given that a substantial amount of clinically relevant information is embedded in unstructured data, natural language processing (NLP) plays an essential role in extracting valuable information that can benefit decision making, administration reporting, and research. Here, we share several desiderata pertaining to development and usage of NLP systems, derived from two decades of experience implementing clinical NLP at the Mayo Clinic, to inform the healthcare AI community. Using a framework, we developed as an example implementation, the desiderata emphasize the importance of a user-friendly platform, efficient collection of domain expert inputs, seamless integration with clinical data, and a highly scalable computing infrastructure.

Prevalence of early repolarization pattern and its association with sudden cardiac death and arrhythmia over one-year follow-up in an Egyptian cohort.

BACKGROUND AND OBJECTIVES: Early repolarization pattern (ERP) is not uncommon electrocardiography (ECG) finding and could be associated with arrhythmia and sudden cardiac death (SCD). We aimed to prospectively determine the prevalence of ERP and its association with arrhythmia and SCD during one-year follow-up in an outpatient Egyptian cohort. METHODS: Clinical assessment and ECG were performed to 1850 consecutive individuals presented at the outpatient clinic of Suez Canal University Hospital (SCUH). Then, the ERP group and 100 age and gender-matched ERP -ve controls had undergone echocardiography, 24-h Holter ECG and exercise stress ECG. RESULTS: ERP was found in 124 individuals (6.7%); we excluded 24 patients with structural heart disease. ERP group (No. = 100) were relatively young (80% <50 years-old) and showed male preponderance (60%). ERP frequencies were: inferolateral (50%), antero-lateral (38%), inferior (10%), and global (2%). ERP subjects were leaner than controls (BMI was 25.3 vs. 30 kg/m2, P value < 0.001) and achieved more metabolic equivalents (METS) on stress ECG (10.7 vs. 8.5 METS, P value < 0.01). Only 4% in the ERP group had horizontal/descending ST slope, while 8% had ST elevation ≥ 2 ms. No arrhythmia or SCD were reported during 1-year follow-up in both groups. Regression analysis showed that male gender, Sokolow-Lyon criteria and short QTc were significant independent predictors of ERP, P value < 0.05. CONCLUSIONS: In outpatient-based Egyptian cohort, the prevalence of ERP was 6.7%, mostly the inferolateral pattern. Our ERP subjects had low-risk clinical and ECG criteria for malignant ERP. Further epidemiological studies are needed to explore the natural history of ERP.