Teaching a knee
to predict
its own future.
ACLART intersects anterior cruciate ligament reconstruction (ACLR) with AI machine learning and in silico simulation. Individual patient clinical & surgery data are fed into ML algorithms and coupled with finite-element computational simulation.
The output is a per-patient prediction of surgical outcome, measured across validated instruments: KOOS, TALS, LKSS, and IKDC-SKF — used both to improve patient selection and to stratify revision-surgery probability.
Deliverable: a point-of-care clinical calculator for ACLR, externally validated on independent datasets, disseminated through peer-reviewed publication and a national SPOT seminar.
Patient outcomes & long-term prognosis
Subjective failure risk at patient-specific level during pre-op discussion.
Refined technique & rehabilitation
Tailored to graft choice, tunnel orientation, fixation device.
Reduce complications & failures
Identify risk factors before the surgical team commits to a plan.
Research platform beyond the grant
Large-scale continuous analysis; protocol for new contributing centres.
Deeper biomechanical insight
Simulations that expose graft / bone / boundary interactions.
Advance knowledge & dissemination
Journals, conferences and a national SPOT seminar.
Clinical data collection
Pre- and post-op records, surgical reports, MRI imaging; Portuguese centres + Scandinavian registers.
Pre-processing & data mining
Outliers, normalization, one-hot encoding, feature selection, dimensionality reduction.
ML modelling
Logistic regression · SVM · random forests · neural networks, benchmarked on KOOS QOL regression.
In silico FEA
FE models of graft/bone/fixation, validated against experimental knee replicas.
Validation & dissemination
External datasets, point-of-care calculator, peer-reviewed publication.
★ Principal investigator. The research team spans surgical medical, AI & ML, and biomodelling sub-groups.