Individual-based modelling
How local cell-level rules generate tissue-level behaviour, and how model structure can make assumptions testable rather than hidden.
Areas I want to explore
Computational models that earn trust through validation.
My interests are intentionally framed as directions for further study, not as claims of participation in a laboratory programme. They align with publicly documented work in Sheffield’s Complex Systems Modelling group: individual-based models, multiscale biomedical modelling, Electrical Impedance Spectroscopy and machine-learning decision support.
Multiscale evidence
How measurements, mechanistic simulation and statistical learning can be connected across scales without confusing model fit with biological explanation.
EIS + machine learning
How signal quality, probe/device variation, cohort composition and calibration affect the validity of decision-support models.
Research questions
Questions I would bring to a project.
These questions reflect my interest in building evidence chains rather than optimising a single headline metric.
DefineWhat is the scientific or decision problem?
MeasureWhat process generated the observations?
ModelWhich assumptions are mechanistic, statistical or convenient?
ValidateDoes performance hold across subjects, devices or settings?
BoundWhat decisions remain unsupported?
Parameter identifiability
When different parameter combinations produce similar outputs, what experiments or priors are needed before parameters can be interpreted biologically?
Sensitivity and uncertainty
Which parameters dominate outcomes, which uncertainties are reducible, and how should uncertainty propagate into a final recommendation?
External validity
Can a model withstand subject-separated testing, temporal change, hardware differences and new populations rather than only a random internal split?
Prospective usefulness
What prospective or workflow-level evidence is needed before a retrospective classifier can support a real decision?
Literature reading
Open the validation frameworkReading beyond the abstract.
Deep-learning EIS analysis (2025)
Contribution: a modern learning approach for EIS-based detection. Questions: patient-level splitting, cohort selection, calibration, class balance, device/site shift and prospective validation.
FLAMEGPU2 cell microenvironment model (2024)
Contribution: scalable agent-based simulation. Questions: whether speed and scale preserve biological fidelity, and how parameters and emergent behaviours are validated against independent observations.
Multicellular neuroblastoma model (2024)
Contribution: mechanistic hypotheses across interacting cell processes. Questions: calibration strategy, global sensitivity, structural alternatives and what evidence would discriminate explanation from curve-fitting.
Virtual tissue and electrical properties (2024)
Contribution: linking tissue structure to simulated measurements. Questions: morphology distributions, parameter uncertainty, measurement realism and validation against held-out physical data.
Public basis for alignment: Dawn Walker’s Sheffield profile, School grant portfolio and Complex Systems Modelling group.