Encode 68-region cortical MRI thickness and CSF tau biomarkers into an 8-qubit ZZFeatureMap Hilbert space — TauNet 1.0 delivers single-patient scoring, batch CSV assessment, and SHAP feature attribution on Kryptur infrastructure.
Recognised among leading quantum-assisted diagnostics platforms, Kryptur advances smarter neurodegenerative intelligence through Hilbert-space kernel methods, CSF biomarker fusion, and live Hetzner VPS API infrastructure.
Single-patient and batch CSV risk scoring with gauge visualization and Chart.js feature attribution.
Eight-qubit encoding with reps=2 and full entanglement — kernel fidelity |⟨φ(xᵢ)|φ(xⱼ)⟩|².
Noise-normalised ibm_fez quantum kernel achieves 3-fold CV AUC 0.9167 vs simulator baseline 0.67.
Authenticated JSON payload accepts 68 cortical values plus APOE, CSF, and MMSE biomarkers.
Per-region cortical thickness parcellation PCA-reduced before quantum statevector assignment.
Partner with quantum-biology experts across kernel SVM design, SHAP attribution, and tier-1 reporting.
Deployed noiseless statevector kernel serves production requests with 3-fold CV AUC 0.67.
Raw QPU kernel validation on IBM Heron r3 achieves AUC 0.75 before noise normalisation.
Noise-normalised quantum kernel reaches AUC 0.9167 — strongest class separation in cohort.
Two hundred synthetic patients with balanced healthy/AD labels train the precomputed kernel SVM.
Flask tier-1 backend serves /api/v1/alzheimers-risk with CORS-enabled batch CSV ingestion.
Optimise neurodegenerative risk outcomes from Hilbert-space kernel and CSF biomarker investments.
Same-origin proxy at /api/taunet/alzheimers-risk forwards JSON patient payloads to the ALZQAPI Flask backend on Hetzner VPS. Request body: 68 cortical thickness values, APOE ε4 count, CSF Aβ42, t-tau, p-tau181, and MMSE. Response includes risk_score (0–1) and top_3_features SHAP attributions.
A quantum kernel support-vector pipeline that encodes patient biomarkers into an 8-qubit Hilbert space and estimates preclinical Alzheimer's risk from state fidelity.
This is an experimental research tool. The risk scores are for demonstration purposes only and should not be used for clinical decision-making.
Vector is PCA-reduced to 8 components before encoding into the ZZFeatureMap.
What's New at Kryptur — TauNet 1.0 API releases, MRI genetics research, and quantum kernel news.