Special Session: Time-series Forecasting with Healthcare Data
Time-series forecasting is a critical challenge in healthcare. Both on an individual level to anticipate patient outcomes and effectively manage disease, and on an institutional level to predict the dynamic demand on services.
In recent years, time-series forecasting has improved significantly by integrating techniques from machine learning (ML) due to their ability to model complex, non-linear patterns in data, unlike traditional statistical methods. Additionally, the integration of increasingly large datasets from diverse sources, such as wearable devices and electronic health records and large repositories of secure data allows for more robust and personalised forecasts.
However significant challenges remain, often stemming from unreliable data quality, highly complex and heterogenous data. Additionally, transparency is essential for these tasks with explainability often being limited for many ML models.
The solution to these challenges will require cooperation between specialists in the medical, mathematical and computer science domains. This special session aims to bring together innovators in these fields to collaborate to maximise the potential for artificial intelligence (AI) and ML to improve time-series forecasting methods to improve patient outcomes and medical resource management.
We encourage contributions from a wide range of disciplines, including biostatistics, medical informatics, computer science and clinical medicine that address both the opportunities and challenges in applying AI and ML to time-series forecasting. This includes any work that seeks to leverage AI and ML systems that conduct time-series forecasting on medical data, or in a medical setting.
The session will highlight topics such as AI-driven resource optimisation, personalised patient care, predictive diagnosis, real-time monitoring and early warning systems.
Topics of Interest (not limited to):
- Time Series Data Modelling: Tackling the unique complexities inherent in medical time-series data analysis.
- Multimodal Data Analysis: Combining diverse data sources to create a holistic understanding of patient health and hospital demand.
- Diagnostic AI Innovations: Enhancing disease identification, particularly for conditions that remain undiagnosed.
- Real-Time Patient Monitoring: Developing AI tools for continuous health tracking and symptom management.
- AI Predictive Models: Crafting and validating AI models to forecast the progression of complex conditions, or hospital demand.
- Comparison of Approaches: Assessing different AI approaches for their effectiveness in the context of medical time-series forecasting.
- Time-Series Foundation Models: Numerous foundation models have been developed for both time-series forecasting and medical diagnosis in a range of settings. We are interested to explore how these foundation models can be applied to improve medical care delivery.
- Explainable AI: Advancing AI systems that prioritise interpretability by both hospital administrators, clinicians and patients.
- Secure AI: Establishing AI protocols that protect patient privacy and data security.
Organisers:
- Matt Ploszajski, Swansea University, 846427@swansea.ac.uk
- Suraj Ramchand, Exeter University, S.Ramchand@exeter.ac.uk
Submission format:
Authors are encouraged to submit original research articles, reviews, or case studies. This Special Session welcomes both full length papers (12 pages plus up to 2 pages of references) and abstracts (up to 5 pages including references). Click here for detailed submission guidelines and templates.
Deadline:
All deadlines, including submission deadline and review timeline are the same as the main conference. Please follow this link to see all the Important Dates.
If you have any questions regarding this Special Session, please contact the organiser.