Logo
EugeneTechRN

Community Hub

Connect with 18,400+ learners — ask questions, share resources, and grow together

PS
Dr. Priya SharmaEducatorHealth Informatics Pinned2h ago

Best practices for de-identifying patient data before ML model training

I've been working on a clinical NLP project and we're hitting some challenges with HIPAA-compliant de-identification. Has anyone used Microsoft Presidio or similar tools in a health informatics pipeline? Would love to hear about your workflows...

#HIPAA#NLP#De-identification
TA
Taiwo AdeyemiLearnerAI & ML4h ago

Confused about when to use LSTM vs Transformer for time-series forecasting

I'm building a model to predict hospital readmission rates over time. I initially went with LSTM but a colleague suggested switching to a Transformer architecture. The course covers both but doesn't directly compare them for this use case...

#LSTM#Transformer#Time Series
NE
Ngozi EzeLearnerData Science6h ago

Share your favourite free datasets for practising data cleaning

I find that the best way to really learn Pandas is to work with genuinely messy real-world data. I'll start: the WHO Global Health Observatory data is incredible for health informatics practice — inconsistent formats, missing values, mixed encodings...

#Datasets#Pandas#Data Cleaning
JO
Jamie OkonkwoEducatorCoding for Kids1d ago

Tips for keeping 10–12 year olds engaged during 45-minute coding sessions

I teach the Scratch and Python for Kids courses here and I'm always looking for engagement strategies. What I've found works: 15-min sprints, mini-challenges with visible output, and letting them pick their own project themes. What works for you?

#Teaching#Engagement#Scratch
RM
Rafael MendesLearnerAI & ML1d ago

My experience deploying the capstone ML model to AWS Lambda — what I learnt

Just completed the ML Fundamentals capstone and wanted to share my deployment experience. I went with AWS Lambda + API Gateway instead of the suggested Flask approach. Cold start times were a real issue with scikit-learn models — here's how I solved it...

#Deployment#AWS#Production
AD
Amara DialloLearnerHealth Informatics2d ago

Understanding HL7 FHIR R4 vs DSTU2 — practical migration considerations

Our hospital is planning to migrate our EHR integration from FHIR DSTU2 to R4. The course covers FHIR basics but doesn't go deep on migration. Has anyone been through this? Particularly interested in how breaking changes in the Patient resource affected your downstream systems...

#FHIR#EHR#Migration
FA
Fatima Al-RashidEducatorData Science3d ago

[Resource] Curated list of open-access health informatics research papers for 2026

I've compiled 40+ freely accessible papers covering EHR analysis, clinical NLP, predictive modelling in ICU settings, and more. All are from 2024–2026 and include implementation code where available. Sharing the full list below...

#Research#Papers#Open Access
KM
Kofi MensahEducatorData Science4d ago

How to interpret a confusion matrix when classes are severely imbalanced

A common mistake I see learners make is optimising for accuracy when their dataset is 95% negative class. This thread covers: why accuracy is misleading, how to read precision/recall/F1, and when to use ROC-AUC vs PR-AUC for imbalanced classification...

#Imbalanced Data#Metrics#Classification