Talks

Title: Constrained Training of Neural Networks via Theorem Proving
Speakers: Mark Chevallier and Matt Whyte
Abstract:

Robotic movement can be trained using neural networks but the process can be lengthy and has no guarantee that safety rules are learned. A neurosymbolic approach can bring the benefits of formal logical constraints specifying safety rules that can be injected into the training process. We introduce our work using logical constraints to assist the training of neural networks via a theorem proving process. In addition to learning via imitation, the neural network evaluates and learns from error caused by breaching these constraints. Our process formally proves the soundness of the logical loss function and guarantees correct implementation of that function using code generation. We discuss our existing work using linear temporal logic to train dynamic movement primitives and go on to discuss future extensions.

Title: ML-based premise selection for Lean
Speaker: Ramon Fernández Mir
Abstract:

In this talk, I will introduce a machine-learning-based tool for the Lean theorem prover that suggests relevant premises to a user interactively constructing a proof. The tool, entirely written in Lean 4, is designed to be highly user-friendly, customizable, and efficient. It is based on a version of random forest, trained on data extracted from mathlib -- Lean's mathematics library. I will discuss one of the main challenges, which was producing useful training features and labels. Finally, I will give a short demo and talk about some interesting related work.

Title: Understanding the Rehabilitation Pathways in Hospital for Patients with Acute COVID-19: A Process Mining Approach
Speaker: Konstantin Georgiev
Abstract:

The delivery of in-hospital rehabilitation during the pandemic presented a serious cause for concern. Patients surviving the infection stage frequently required prolonged and more complex treatment. Thus, rehabilitation wards have undergone rapid changes in the delivery of these services. For the first time, detailed contact data with associated timestamps and activities performed by Allied Healthcare Professionals (AHPs) has become routinely available in Electronic Health Records. This data could be key to understanding complex rehabilitation pathways and identifying areas for improvement in COVID-19, resulting in better clinical outcomes. In this study, I will explore the sequences of AHP-related activities in patients presenting with COVID-19 across three acute hospitals within NHS Lothian. I will try to assess the treatment efficiency of the delivered services between the first and second waves using process maps and metrics relative to recovery time.

Title: Stochastic block modelling and link prediction to improve mortality prediction for critical patients.
Speaker: Jorge Gaete Villegas
Abstract:

Mortality prediction for patients in Intensive Care Units (ICU) is an important but challenging task. Early prediction can improve medical outcomes, optimize medical interventions, and minimize the use of resources. Current efforts to create mortality prediction models rely on medical consensus, regression methods, and machine learning. Unfortunately, the nature and quality of  ICU data can affect the performance of such models. Some of the shortcomings reported in the literature include the overestimation of mortality for older patients and low predictive power for underrepresented patient groups. In this talk we present our current work exploring stochastic block modelling and link prediction to forecast mortality and overcome such shortcomings.

Title: What do data tell us about frailty?
Speaker: Lara Johnson
Abstract:

My research integrates data science and geriatric medicine to explore what different data sources and methods can tell us about frailty, a state of increased vulnerability to adverse health outcomes for individuals of the same chronological age. I am looking at how the number and combination of health issues people have – a proxy measure for frailty - relate to their functioning ability (such as their ability to make a cup of tea, get dressed or walk up a flight of stairs) and adverse health outcomes (death, falls, fractures, care needs).  This will inform the development of a data-driven definition of frailty (currently lacking), which is useful both for identifying patients in later life at highest risk as well as forecasting demand on health and social care services. My research aims to answer questions such as whether there are different types of frailty, whether frailty manifests differently in men vs. women (who are more frail but live longer) and whether distinguishing between types of frailty improves the performance of prediction models.

Title: Human activity recognition and Identifying the activity patterns on noninvasive sensor data with Deep Learning
Speaker: Simon U
Abstract:
The recent advancement and development of both embedded electronic devices and deep learning techniques have made real-time activity tracking and monitoring feasible with the help of wearable devices or cameras. The time-series data can be analyzed and used to track an individual's health condition and daily activities routine. But there's a growing trend toward using noninvasive and non-visual activity sensing to get information and figure out what a person is doing without bothering them. No one wants to be constantly watched and recorded by cameras.

The current approach for human activity recognition using ambient sensors, such as motion sensors and light sensors, is restrictive and has poorer performance compared to approaches using cameras and wearable sensors. In this talk, I will discuss the challenges encountered on the task as well as potential approaches I hope to investigate in order to overcome some of these issues during my MInf project. The aim is to build a model that takes into account prior knowledge of common activity patterns for the activity recognition task and uses the same model to forecast the individualized activity routine.

Title: Creating a GUI for the Proter simulator
Speaker: Gareth Dawson
Abstract:

Since it's original development for Digiflow, Proter has evolved into a general-purpose business process simulator. In a previous project a scala based web server and restful API was created for Proter. We will use this as the backend for a graphical user interface. The GUI should expose the existing core functionality which is currently only available programmatically. The GUI should also be well designed and evaluated for usability and functionality.

Title: Detecting long-term deviation in Activities of Daily Living based on sensor data
Speaker: Leo Kravtchin
Abstract:

Activities of Daily Living behaviour patterns and daily routines, which include any activities performed on a daily basis, can give many insights into a person's mental and physical state. This project is based on the already existing CASAS datasets and their developed supervised human activity recognition algorithm to label raw time series sensor data with the performed activity, based on many ambient sensors' readings in a participant's home. However, as one might expect, not all activities are classified correctly by the developed algorithm. This talk discusses analysis of the existing data, as well as already conducted and planned experiments to improve the algorithm's performance by combining hierarchical classification and deep learning techniques. Also, some of the next planned steps are explored, such as the detection of correlations between external weather factors and a person's daily routine deviations to draw conclusions about the impact of such events, with an emphasis on mobility-related ADL patterns. Examples of this can be wandering around at unusual times or a change in the duration of outside activities, depending on seasonality and weather changes. Finally, the realism of the original aim of this project to detect long-term deviations and health deterioration based on the CASAS data is discussed.

Title: Bayesian inference of disease co-morbidity networks from electronic health records
Speaker: Guillermo Romero Moreno
Abstract:

Co-morbidity networks (a.k.a. Phenotypic disease networks) capture associations between morbidities that can be later used to explore disease progression, find typical co-morbidity patterns, etc. However, current methods for building co-morbidity networks suffer known biases and their statistical properties are not well defined or exploited. Here, we provide a new method for building comorbidity networks that correct some of the previous biases by disentangling the effects of co-morbidity and the ‘natural’ mechanisms for a morbidity to appear independently. This is achieved via a Bayesian approach that infers the latent target quantities from the data while also providing full statistical distributions of the inferred quantities and other network metrics. The Bayesian approach provides further benefits since i) it is based on a flexible and interpretable model that can be understood by domain experts; ii) it allows to inject domain knowledge in the models and priors; iii) it can generate alternative realisations of the original data, which can be particularly useful for comparing sub-populations at the level of raw co-occurrence of diseases. We apply this methodology to a primary care dataset and explore the benefits and uses of this methodology, while also comparing it to previous methods for building co-morbidity networks, such as relative risk or φ-correlations. We show that our methodology provides a more robust quantification of risk and that our results are more amenable to statistical comparison.

Title: Networks for smoking dynamics
Speaker: Adarsh Prabhakaran
Abstract:

Tobacco usage, one of the leading public health issues globally, has declined over the years due to multiple policies. Despite these policies, the decline of the number of smokers is slowing down, and tobacco use is still common. Over the years, multiple models have been developed to study the spread of smoking and to develop tobacco control strategies (mainly ODE - Ordinary Differential Equation models). However, these models do not consider all interactions between individuals observed empirically and their underlying network structure. In this talk, I will describe the network-based Agent-based model we developed to reproduce the social contagion process of tobacco use. Our results suggest that network structure is essential and that the observed dynamics from the ODE model are only similar to the network-based ABM only when the underlying network is fully connected, which is rarely the case. Further, by comparing multiple theoretical networks, we also show that networks with a similar average degree as the real-world network can be used to model smoking behaviour when complete information on the real-world network is unavailable. Additionally, our ABM can be used as a testbed to develop network-based intervention strategies and policies for tobacco control.