PhD Project: Improving Discharge Care Coordination in Hospitals using Machine Learning and Probabilistic Data-Driven Approaches

Kurdish Annual Academic Conference 2026

Mustafa Aslan

Data Lab for Social Good, Cardiff University, UK

Lead supervisor: Prof. Bahman Rostami-Tabar
Co-supervisor: Dr. Jeremy Dixon

7 October 2026

Outline

Who am I?
My PhD Project
Project: Understanding Causal Impact of Diagnosis
Project: From Forecasting Competition to Impact in NHS Bristol

Outline

Who am I?
My PhD Project
Project: Understanding Causal Impact of Diagnosis
Project: From Forecasting Competition to Impact in NHS Bristol

My Journey

Born: Xirabreşk - Close to Girê Mirazan (Göbekli Tepe), Urfa, Turkey

Education: 📚
→ Moved to Ankara for university in 2010
→ BSc in Business Administration (METU, 2015)
→ MSc in Financial Mathematics (METU, 2021)

Industry (9 years of experience): 💼
→ Worked while completing Master’s degree
→ Built expertise in forecasting, ML, and data analytics

Current: 🎓
→ PhD at Cardiff University (Oct 2024 – Present)
- Focus: Probabilistic Modeling in Healthcare
- WGSSS-ESRC Studentship Award recipient

My CV at a Glance

🎓 Education

PhD in Probabilistic Modeling in Healthcare
Cardiff University, UK | Oct 2024 – Present
✓ WGSSS-ESRC Studentship Award

MSc in Financial Mathematics
METU, Turkey | 2017–2021

BSc in Business Administration
METU, Turkey | 2011–2015

💼 Work Experience

Data Analytics Executive | AKBANK, Istanbul
Jun 2022 – Sep 2024
→ Predictive modeling & forecasting

Research Associate | Economic Policy Research Foundation
Jan 2022 – Apr 2024
→ Statistical inference & modeling

Senior Process Analyst | ETI GIDA
Jun 2019 – Jan 2022
→ ERP & BI tools, process optimization

🏆 Achievements & Awards

SPHERE-PPL Forecasting Contest (2026)
🥇 1st Place (1–5 day horizon)
🥈 2nd Place (5–10 day horizon)
→ Predicting avoidable NHS hospital deaths

Best Presentation Award
5th Welsh Postgraduate Research Cluster
Workshop (Sep 2025)

🛠️ Skills & Expertise

Expertise: ML, Time Series Forecasting, Probabilistic Forecasting, Stochastic Optimization, Statistical Modeling

Programming: Python, SQL

Languages: English (Fluent) • Kurdish (Mother Tongue) • Turkish (Native)

Outline

Who am I?
My PhD Project
Project: Understanding Causal Impact of Diagnosis
Project: From Forecasting Competition to Impact in NHS Bristol

Very Brief Overview

Project 1: Staffing Required Number of Nurses in a Mental Health Hospital using Probabilistic Forecasting and Optimization

Project 2: From Causal Impacts of Diagnosis to Reclaimed Hospital Capacity using Causal AI

Project 3: Forecasting Daily Number of Hospital Admissions&Discharges per Specialty (40+ Specialties) using ONE Forecasting Model to Improve Hospital Capacity Planning and Resource Allocation

Outline

Who am I?
My PhD Project
Project: Understanding Causal Impact of Diagnosis
Project: From Forecasting Competition to Impact in NHS Bristol

Patients Come to Hospital with a Diagnosis (Reason) and What is the Impact of Diagnosis on Length of Stay (LoS)?

Aim: To quantify the causal impact of diagnosis on Length of Stay (LoS) in Hospital using Causal AI.

Outline

Who am I?
My PhD Project
Project: Understanding Causal Impact of Diagnosis
Project: From Forecasting Competition to Impact in NHS Bristol

Winner • Joint 1st SPHERE-PPL Forecasting Contest • NHS Acute Patient Harm (1–5 Day Category)

2nd Place (After Microsoft) SPHERE-PPL Forecasting Contest • NHS Acute Patient Harm (5–10 Day Category)

Forecasting Severe Acute Patient Harm Resulting from Emergency Department Delays in the NHS Bristol

Winning Team • Data Lab for Social Good

Prof. Bahman Rostami-Tabar

Professor of Analytics & Decision Sciences

Director, Data Lab for Social Good Research Group
Cardiff Business School, Cardiff University

Mustafa Aslan

PhD Student

Data Lab for Social Good Research Group
Cardiff Business School, Cardiff University

Harsha Halgamuwe Hewage

PhD Student

Data Lab for Social Good Research Group
Cardiff Business School, Cardiff University

Organized by SPHERE (SPatial, Health & Environmental REsearch)

The Real Problem 🏥

Emergency Department Delays & Patient Lives

When patients wait for hospital beds:

Risk of harm increases significantly
~25 preventable deaths/month in Bristol NHS due to waiting
Staff need early warning to prevent these delays

The Competition Challenge 📊

Predict hospital harm 10 days in advance

Using: Real hospital data from Bristol NHS
Goal: Give hospital managers early warning so they can take pre-emptive action
Data source: 24 months of real hospital records

Forecast Evaluation: Future 120-Days (example)

Note: Forecasts are generated once every 5 days, 24 times across 120 days.

Can Point Forecasts Capture Peak Dynamics?

Any questions or thoughts? 💬