BACK
DEC 10, 2025
Health Risk Assessment
Over the last few weeks, my team and I took a raw clinical dataset and turned it into a fully functional health risk assessment web app as a project of our Machine Learning course.
Our tool is motivated by the NEWS2 (National Early Warning Score 2) protocol. However, while standard protocols rely on manual scoring, our tool leverages ML algorithms to classify risk levels automatically.
We didn't just want to train a model, but we wanted to build a complete end-to-end pipeline. We started with a dataset of vital signs and clinical indicators (RR, SpOโ, BP, etc.) and compared multiple algorithms, including SVM, KNN, and Random Forest, to find the best one.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฐ๐ต ๐ฆ๐๐ฎ๐ฐ๐ธ & ๐ช๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐:
๐ญ- ๐๐ฎ๐๐ฎ & ๐ ๐ผ๐ฑ๐ฒ๐น๐ถ๐ป๐ด: Scikit-learn Pipelines with a focus on data leakage prevention.
๐ฎ- ๐จ๐/๐จ๐ซ: Designed in Figma and deployed in Streamlit.
๐ฏ- ๐๐ฒ๐ฎ๐๐๐ฟ๐ฒ๐: Real-time risk classification, PDF report generation, and an encryption demo to highlight privacy.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฎ๐บ ๐๐ณ๐ณ๐ผ๐ฟ๐:
I had the pleasure of leading this project with Jehad Albarak and Ahmed Al naim.
- Jehad Albarak: led the deep dive into SVMs and PCA analysis to understand our feature space.
- Ahmed Al naim: optimized our KNN implementation and handled the comprehensive metrics evaluation.
- ๐ ๐ ๐๐ผ๐ฐ๐๐: I managed the pipeline architecture, notebook organization and documentation, the Random Forest implementation, and the deployment by taking our code from a Colab notebook to the live web app.
๐ง๐ฟ๐ ๐ถ๐:
https://lnkd.in/dMQ9J_w3
๐๐น๐ฎ๐ฟ๐ถ๐ณ๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป: This is a prototype tool for AI-assisted health risk classification based on vital signs and clinical indicators. It is primarily designed for educational and research use only.
Mohammed Aldosari
AI Engineer ยท Riyadh
#MachineLearning #Streamlit #DataScience #Python #HealthTech #AI #kaggle
DEC 10, 2025
Health Risk Assessment
Over the last few weeks, my team and I took a raw clinical dataset and turned it into a fully functional health risk assessment web app as a project of our Machine Learning course.
Our tool is motivated by the NEWS2 (National Early Warning Score 2) protocol. However, while standard protocols rely on manual scoring, our tool leverages ML algorithms to classify risk levels automatically.
We didn't just want to train a model, but we wanted to build a complete end-to-end pipeline. We started with a dataset of vital signs and clinical indicators (RR, SpOโ, BP, etc.) and compared multiple algorithms, including SVM, KNN, and Random Forest, to find the best one.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฐ๐ต ๐ฆ๐๐ฎ๐ฐ๐ธ & ๐ช๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐:
๐ญ- ๐๐ฎ๐๐ฎ & ๐ ๐ผ๐ฑ๐ฒ๐น๐ถ๐ป๐ด: Scikit-learn Pipelines with a focus on data leakage prevention.
๐ฎ- ๐จ๐/๐จ๐ซ: Designed in Figma and deployed in Streamlit.
๐ฏ- ๐๐ฒ๐ฎ๐๐๐ฟ๐ฒ๐: Real-time risk classification, PDF report generation, and an encryption demo to highlight privacy.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฎ๐บ ๐๐ณ๐ณ๐ผ๐ฟ๐:
I had the pleasure of leading this project with Jehad Albarak and Ahmed Al naim.
- Jehad Albarak: led the deep dive into SVMs and PCA analysis to understand our feature space.
- Ahmed Al naim: optimized our KNN implementation and handled the comprehensive metrics evaluation.
- ๐ ๐ ๐๐ผ๐ฐ๐๐: I managed the pipeline architecture, notebook organization and documentation, the Random Forest implementation, and the deployment by taking our code from a Colab notebook to the live web app.
๐ง๐ฟ๐ ๐ถ๐:
https://lnkd.in/dMQ9J_w3
๐๐น๐ฎ๐ฟ๐ถ๐ณ๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป: This is a prototype tool for AI-assisted health risk classification based on vital signs and clinical indicators. It is primarily designed for educational and research use only.
Mohammed Aldosari
AI Engineer ยท Riyadh
#MachineLearning #Streamlit #DataScience #Python #HealthTech #AI #kaggle
DEC 10, 2025
Health Risk Assessment
Over the last few weeks, my team and I took a raw clinical dataset and turned it into a fully functional health risk assessment web app as a project of our Machine Learning course.
Our tool is motivated by the NEWS2 (National Early Warning Score 2) protocol. However, while standard protocols rely on manual scoring, our tool leverages ML algorithms to classify risk levels automatically.
We didn't just want to train a model, but we wanted to build a complete end-to-end pipeline. We started with a dataset of vital signs and clinical indicators (RR, SpOโ, BP, etc.) and compared multiple algorithms, including SVM, KNN, and Random Forest, to find the best one.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฐ๐ต ๐ฆ๐๐ฎ๐ฐ๐ธ & ๐ช๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐:
๐ญ- ๐๐ฎ๐๐ฎ & ๐ ๐ผ๐ฑ๐ฒ๐น๐ถ๐ป๐ด: Scikit-learn Pipelines with a focus on data leakage prevention.
๐ฎ- ๐จ๐/๐จ๐ซ: Designed in Figma and deployed in Streamlit.
๐ฏ- ๐๐ฒ๐ฎ๐๐๐ฟ๐ฒ๐: Real-time risk classification, PDF report generation, and an encryption demo to highlight privacy.
๐ง๐ต๐ฒ ๐ง๐ฒ๐ฎ๐บ ๐๐ณ๐ณ๐ผ๐ฟ๐:
I had the pleasure of leading this project with Jehad Albarak and Ahmed Al naim.
- Jehad Albarak: led the deep dive into SVMs and PCA analysis to understand our feature space.
- Ahmed Al naim: optimized our KNN implementation and handled the comprehensive metrics evaluation.
- ๐ ๐ ๐๐ผ๐ฐ๐๐: I managed the pipeline architecture, notebook organization and documentation, the Random Forest implementation, and the deployment by taking our code from a Colab notebook to the live web app.
๐ง๐ฟ๐ ๐ถ๐:
https://lnkd.in/dMQ9J_w3
๐๐น๐ฎ๐ฟ๐ถ๐ณ๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป: This is a prototype tool for AI-assisted health risk classification based on vital signs and clinical indicators. It is primarily designed for educational and research use only.
Mohammed Aldosari
AI Engineer ยท Riyadh
#MachineLearning #Streamlit #DataScience #Python #HealthTech #AI #kaggle