Selected Research & Consulting Work
Examples of projects across health research, statistical analysis, evidence synthesis, mixed methods, programme evaluation and advanced analytical research.
Some project details have been anonymised or generalised to protect client confidentiality.
Clinical Data & AI
Designing Reliable Clinical Data Systems for AI-Ready Decision-Making
Development of a structured interpretation framework for complex mental health data, improving consistency and preparing the data for reliable analytical and AI use.
Clinical data · Data frameworks · AI readiness
Clinical Statistics
Statistical Analysis of Health Data for Clinical Research
Analysis of complex patient data using structured data preparation, statistical testing and regression modelling to support clinical interpretation and publication.
Clinical data · Regression modelling · Publication support
Evidence Synthesis
Systematic Review & Meta-Analysis for Defensible Clinical Evidence
Structured evidence synthesis combining systematic review methodology, critical appraisal and meta-analysis to support clear, publication ready clinical conclusions.
Systematic review · Meta-analysis · Evidence synthesis
Programme Evaluation
Evaluating a Public Health Programme Using Mixed Methods
Mixed-methods evaluation combining health outcomes and contextual evidence to assess programme effectiveness and support decisions around implementation and scale.
Mixed methods · Health outcomes · Programme evaluation
Additional Applied Research
QUALITATIVE RESEARCH & THEMATIC ANALYSIS
Qualitative Analysis & Thematic Coding for a Multi-Case Strategy Research Project
Systematic analysis of complex interview data using thematic coding, codebook development and NVivo-supported analysis to create a structured and traceable qualitative evidence base.
Qualitative analysis · Thematic analysis · NVivo · Coding frameworks
RESEARCH DESIGN & METHODOLOGY
Research Design & Methodology for a Multiple Site Value Chain Assessment
A complex international development research programme was translated into a structured mixed-method framework covering sampling, survey design, KoBoToolbox implementation, variable definition and analytical planning.
Mixed methods · Survey design · Sampling · KoBoToolbox
Featured Research & Publication
Understanding Social Support in Online Diabetes Communities
Mixed-Methods Research · Machine Learning · Health InformaticsExploring how individuals seek and provide social support in an online diabetes community, taking into consideration users' progressive disease state by using advanced analytical methods. Read the publication
Research Challenge in Online Health Data Analysis
Online health communities have rich, unstructured text data on patients experience. The challenge was to understand how different types of support were expressed, how support varied across stages of the condition, and how these interactions could inform research and practical interventions.
Mixed-Methods and Machine Learning Approach
Development of a mixed-methods analytical framework that combined qualitative research with supervised machine learning techniques to retain depth of insight while enabling scalable analysis to identify, quantify, and compare patterns of social support.
Statistical Analysis and Machine Learning Implementation
- Defined and refined social support categories through qualitative analysis
- Built and annotated a training dataset
- Engineered linguistic and behavioural features
- Developed and evaluated machine learning models
- Assessed patterns across diabetes stages
1. Enabled large scale analysis of social support exchanges in an online diabetes community.
2. Revealed patterns in how individuals seek and provide support across diabetes progressive stages.
3. Provided a framework applicable to other chronic conditions and online health communities.
Key Insight
Online diabetes communities function as dynamic ecosystems of social support, where users can move between seeking support and sharing knowledge as their experiences and needs evolve. When analysed effectively, these interactions reveal how support needs differ across individuals and contexts — providing valuable insight to inform more targeted, data-driven patient support and engagement strategies.
User Data
Posts
Feature Extraction
Text signals
Model
Classification
Output
Support types
Transforming unstructured data into reliable, decision-ready insight.
Published in peer-reviewed journal.
Discuss Your Project
Whether you need support with research design, analysis, evidence synthesis, evaluation or publication, Wells can provide specialist input at the stage you need it.