Conformalised data synthesis
Authors:
Julia A. Meister,
Khuong An Nguyen
Abstract:
With the proliferation of increasingly complicated Deep Learning architectures, data synthesis is a highly promising technique to address the demand of data-hungry models. However, reliably assessing the quality of a 'synthesiser' model's output is an open research question with significant associated risks for high-stake domains. To address this challenge, we propose a unique synthesis algorithm…
▽ More
With the proliferation of increasingly complicated Deep Learning architectures, data synthesis is a highly promising technique to address the demand of data-hungry models. However, reliably assessing the quality of a 'synthesiser' model's output is an open research question with significant associated risks for high-stake domains. To address this challenge, we propose a unique synthesis algorithm that generates data from high-confidence feature space regions based on the Conformal Prediction framework. We support our proposed algorithm with a comprehensive exploration of the core parameter's influence, an in-depth discussion of practical advice, and an extensive empirical evaluation of five benchmark datasets. To show our approach's versatility on ubiquitous real-world challenges, the datasets were carefully selected for their variety of difficult characteristics: low sample count, class imbalance, and non-separability. In all trials, training sets extended with our confident synthesised data performed at least as well as the original set and frequently significantly improved Deep Learning performance by up to 61 percentage points F1-score.
△ Less
Submitted 10 January, 2025; v1 submitted 14 December, 2023;
originally announced December 2023.
Assessing long-term medical remanufacturing emissions with Life Cycle Analysis
Authors:
Julia A. Meister,
Jack Sharp,
and Yan Wang,
Khuong An Nguyen
Abstract:
The unsustainable take-make-dispose linear economy prevalent in healthcare contributes 4.4% to global Greenhouse Gas emissions. A popular but not yet widely-embraced solution is to remanufacture common single-use medical devices like electrophysiology catheters, significantly extending their lifetimes by enabling a circular life cycle. To support the adoption of catheter remanufacturing, we propos…
▽ More
The unsustainable take-make-dispose linear economy prevalent in healthcare contributes 4.4% to global Greenhouse Gas emissions. A popular but not yet widely-embraced solution is to remanufacture common single-use medical devices like electrophysiology catheters, significantly extending their lifetimes by enabling a circular life cycle. To support the adoption of catheter remanufacturing, we propose a comprehensive emission framework and carry out a holistic evaluation of virgin manufactured and remanufactured carbon emissions with Life Cycle Analysis (LCA). We followed ISO modelling standards and NHS reporting guidelines to ensure industry relevance. We conclude that remanufacturing may lead to a reduction of up to 60% per turn (-1.92 kg CO2eq, burden-free) and 57% per life (-1.87 kg CO2eq, burdened). Our extensive sensitivity analysis and industry-informed buy-back scheme simulation revealed long-term emission reductions of up to 48% per remanufactured catheter life (-1.73 kg CO2eq). Our comprehensive results encourage the adoption of electrophysiology catheter remanufacturing, and highlight the importance of estimating long-term emissions in addition to traditional emission metrics.
△ Less
Submitted 9 February, 2023; v1 submitted 28 November, 2022;
originally announced November 2022.
