Approved Research

Deep phenotyping for common and rare diseases

Lay summary

Our research project aims to bring artificial intelligence (AI) and genomics together to transform how we understand and treat diseases. In the world of medicine, understanding diseases and their underlying causes is crucial for creating effective treatments. This understanding often comes from examining patients' genetic information and manually curated sets of diagnostic codes from linked health records. However, the current method of doing this is prone to errors and doesn't take advantage of all the complex data that we can now gather.

In the modern clinical context, a wealth of data beyond traditional health records is available. This includes 'Omics' data (proteomics, metabolomics, etc.), which provides a detailed view of a patient's biological makeup, as well as imaging data (like MRIs and retinal fundus images) and functional test results (like ECGs). But, currently, we don't have efficient ways to use all this complex data in understanding diseases.

That's where our research comes in. We aim to develop sophisticated AI methods to draw meaningful conclusions from this wealth of complex data. Using deep learning, we'll train models to find disease-relevant information from all this data, helping us to develop a more precise and detailed picture of different diseases - we call this 'deep phenotyping'.

Once these AI models are trained, we'll examine the extracted data to understand the similarities and differences across various types of data. We'll then use this refined and combined data to improve our understanding of the genetic architecture of various diseases. The goal is to enable a more nuanced understanding of diseases and their genetic underpinnings.

The project is expected to last for three years. By the end of this period, we aim to have a well-established AI framework capable of harnessing complex biomedical data to produce deep disease phenotypes. These phenotypes will be used to augment the available genetic evidence, enhancing our understanding of diseases and potentially opening up new avenues in drug target discovery.

The potential impact on public health is substantial. By enhancing our understanding of disease mechanisms, our research could revolutionize the way diseases are characterized and treated. This could lead to more accurate diagnoses, more effective treatments, and ultimately, improved health outcomes for patients. Furthermore, our project could also pave the way for a more personalized approach to healthcare, with interventions tailored to the specific disease profiles of individual patients