Mathematical Modelling of Cancer – Adipocyte Interactions in Ovarian Cancer Modelling and Optimising Multi-Modal Customer Journeys in Commercial Banking: A Human-Centred Approach to Behavioural Analytics and Outcome Prediction Across Digital, Physical, and Telephony Channels

—

by

in

Project Overview

This research investigates the use of Human-Centred AI to model and optimise customer journeys in UK commercial banking. The project focuses on multi-modal customer interactions across digital, physical, and telephony channels, such as mobile apps, online platforms, branch visits, and call centre experiences.

Using topic modelling techniques on unstructured customer feedback (including surveys, call transcripts, and chat logs), the project extracts behavioural signals to identify friction points, service gaps, and journey bottlenecks. By combining traditional LDA with emerging interpretability tools such as Gemini-based topic attribution, the research aims to make model outputs accessible to non-technical stakeholders in finance.

Working under the constraints of a secure enterprise infrastructure, the project also explores strategies for deploying NLP tools in environments with limited external connectivity. Ultimately, the goal is to deliver ethically grounded, transparent, and actionable insights that support customer experience optimisation and drive improved Net Promoter Scores (NPS) across service channels.

Project Aims

  • Develop a human-centred framework for modelling and optimising customer journeys in UK commercial banking, incorporating behavioural data across digital, physical, and telephony touchpoints.
  • Design and evaluate methods for analysing unstructured customer feedback, using NLP and topic modelling to extract insights relevant to customer satisfaction and service design.
  • Integrate behavioural analytics across multiple banking channels to identify friction points, unmet needs, and cross-channel interactions that influence customer outcomes.
  • Address deployment constraints by developing workarounds for applying machine learning within secure enterprise infrastructures, including restricted data and implementation environments.
  • Enhance the interpretability and actionability of AI outputs, ensuring stakeholders can use insights ethically and effectively to improve inclusivity and customer experience.

Project Team

Collaborators

Swansea University, HSBC UK – Commercial Banking (CMB) Division