Quantum Twin Learning: A Hybrid Quantum-Classical Methodology for Early Prediction of Neurodegenerative Disease Progression using Multimodal Patient Digital Twins

Neurodegenerative disorders progress silently for up to two decades before diagnosis, yet 
clinical prediction tools remain snapshot-based, single-modality and population-averaged, so forecasting 
begins only after irreversible neuronal loss. This paper proposes Quantum Twin Learning (QTL), a 
methodology that couples a continuously re-synchronised multimodal patient digital twin with a hybrid 
quantum-classical learner to estimate patient-specific progression trajectories inside the preclinical 
window. The twin maintains a time-aligned latent state per patient, built by a fusion layer that harmonises 
imaging, health records, genomics, wearable telemetry and speech across sampling rates and tolerates 
missing modalities. A compact variational quantum circuit, fed by a classical encoder, operates on the twin 
state rather than on raw scans and learns cross-modal correlations with few trainable parameters. Three 
prediction heads output a staged trajectory, a time-to-event risk horizon and calibrated uncertainty, 
trained with a joint objective and updated each time a new observation arrives. An evaluation protocol on 
public longitudinal cohorts, with classical baselines and clinically meaningful metrics, is specified so that 
the method can be validated without hardware-dependent claims. The methodology is designed for 
hospital triage, remote monitoring and precision therapy timing in next-generation smart healthcare 
systems.