There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Elaheh Kalantari |
| Institution: | University of Surrey |
| Department: | Electrical Engineering |
| Country: | |
| Proposed Analysis: | My PhD focus is on machine learning models to analyse in-home observations, measurements and clinical data for dementia patients monitoring. Understanding the basis and trajectories of health and disease need to develop models based on time series and longitudinal data. Analysis of large-scale time series data collected from diverse devices has created new multi-faceted challenges and opportunities; multivariate time series analysis, anomaly detection, forecasting, and imputation are some of the headlines that associated with patient data monitoring. Furthermore, with the healthcare industry becoming increasingly digitized, the amount of data that is generated is growing precipitously. Due to the scale and rate with which this growth is taking place, annotating all this data, to feed into a deep-learning algorithm for instance, is infeasible. In general, patient data can be associated with diagnosis information recorded by clinicians, including comorbidities which lead to various symptoms and changes in patient's health. These associations between diseases and symptoms in health records can cause significant confusion between correlation and causation for automated algorithms. For instance, in case of patients with dementia, sleep disruption may cause sooner occurrence or may worsen the severity of expected symptoms and disorders such as cognitive impairment, but it is challenging to decide whether sleep disruption is a symptom of dementia disease or is a comorbidity case. To address the above-mentioned challenges, this research will focus on machine learning and deep learnings algorithms to i) fuse and represent multimodal data (EHR, image, video) to extract more relevant information that results in more discrepancy between subjects, their disease stages, and more accurate models for generalisability ii) overcome limited access to the high-quality labels in a clinical setting where physicians are squeezed for time and attention, self-supervised representation learning techniques have emerged as promising methods iii) combine logical description methods with analytical methods (to avoid bias and confusion between causation and correlation) iv) create a mapping between the various data points to follow all the temporal/sequential events and patterns v) provide prediction models for the onset of a specific disease, related events such as accelerated disease progression and for disease risk assessment vi) develop graph analytics, network embedding and graph neural networks, to help with the understanding the mechanism, intervention and prevention vii) investigate the relationships among co-occurring symptoms and consequences on longitudinal data for causal inferences and viii) employ continual learning machine learning models for predictive analytics to optimise clinical management decisions in real-time. Therefore, I need data to be able to satisfy the goal of this thesis by considering some conditions; they should be related to the dementia patients, have several types of data for multi-modal approach, they should cover a longitudinal study for prognostic prediction, and causal inference. Besides genetic aspects, I will consider other factors which play substantial roles in dementia onset and progress such as ageing. Based on my studies ADNI, AIBL, and ADNIDOD datasets are the most well-known online data which I have found through recently published papers and I hope I could invest my PhD research focus on these precious datasets. Having access to these datasets will be a fantastic start point for my PhD and help me to develop clinically useful tools that can work on real challenges. |
| Additional Investigators | |
| Investigator's Name: | Samaneh Kouchaki |
| Proposed Analysis: | She is my supervisor, therefore, please consider my proposal. |
