AIAI Seminar - 22 May 2023 - Talks by Zonglin Ji, Andreas Bueff and Claire Barale

 

Speaker:     Zonglin Ji

Title:            In-Hospital Mortality Prediction for ICU Patients with Liver Disease using a Process Mining/Deep Learning Architecture

Abstract: 

Patients with liver disease in the Intensive Care Unit (ICU) present a unique clinical challenge with increased risks of complications and mortality. Their medical management is complex due to the multifaceted nature of the disease and the interplay of various risk factors. Traditional severity scoring methods, while helpful, do not comprehensively consider patients' prior medical history and the time dimension of their disease progression. This research aims to consider such timely information with a process mining/deep learning architecture to enhance the predictive capacity of existing severity scoring methods by incorporating the medical history of liver disease patients.

In this approach, health records of past hospital encounters are transformed into event logs, thereby making them amenable to process mining. These event logs inform the creation of a process model detailing the patients' past hospital encounters. A process mining pipeline is being utilised to combine demographic, medical, and time-series data with established severity scores to predict in-hospital mortality for liver disease ICU patients. The eventual goal of this on-going research is to enable healthcare providers to identify and potentially mitigate risk factors more effectively, thereby improving the outcomes for patients. 

 

Speaker:     Andreas Bueff

Title:            Deep Inductive Logic Programming meets Reinforcement Learning

Abstract: 

One approach to explaining the hierarchical levels of understanding within a machine learning model is the symbolic method of inductive logic programming (ILP), which is data efficient and capable of learning first-order logic rules that can entail data behaviour. A differentiable extension to ILP, so-called differentiable Neural Logic (dNL) networks, are able to learn Boolean functions as their neural architecture includes symbolic reasoning. We propose an application of dNL in the field of Relational Reinforcement Learning (RRL) to address dynamic continuous environments. This represents an extension of previous work in applying dNL-based ILP in RRL settings, as our proposed model updates the architecture to enable it to solve problems in continuous RL environments. The goal of this research is to improve upon current ILP methods for use in RRL by incorporating non-linear continuous predicates, allowing RRL agents to reason and make decisions in dynamic and continuous environments.

 

Speaker:     Claire Barale

Title:             Automated Refugee Case Analysis: A NLP Pipeline for Supporting Legal Practitioner

Abstract:

In this talk, I present an end-to-end pipeline for retrieving, processing, and extracting targeted information from legal cases. We investigate an under-studied legal domain with a case study on refugee law in Canada.

Searching case law for past similar cases is a key part of legal work for both lawyers and judges, the potential end-users of our prototype.

While traditional named-entity recognition labels such as dates are meaningful information in law, we propose to extend existing models and retrieve a total of 19 categories of items from refugee cases.

After creating a novel data set of cases, we perform information extraction based on state-of-the-art neural named-entity recognition (NER).

We test different architectures including two transformer models, using contextual and non-contextual embeddings, and compare general purpose versus domain-specific pre-training.

The results demonstrate that models pre-trained on legal data perform best despite their smaller size, suggesting that domain-matching had a larger effect than network architecture. We achieve a F1 score superior to 90% on five of the targeted categories and superior to 80% on an additional 4 categories.