Reserve estimation is a fundamental actuarial task that underpins premium pricing, solvency assessment, financial reporting, capital planning and risk management. Classical methods based on Thiele’s differential equation offer a rigorous and interpretable life‑insurance valuation, yet repeated calculations become costly in sensitivity analysis, optimization and large‑scale scenario evaluation. This paper introduces an intelligence platform for term‑life reserve modelling that couples a traditional Thiele solver with a physics‑informed neural network (PINN) enhanced by knowledge‑informed neural network (KINN) losses. The framework comprises synthetic policy generation, risk‑adjusted premium calculation, classical reserve trajectory generation, reserve‑ratio dataset construction, configurable neural training, validation diagnostics, sensitivity and elasticity analysis, prototype optimization workflows and interest‑rate scenario testing. A key refinement is the use of premium ratio and the explicit separation of pricing‑time and scenario‑time interest‑rate semantics. The final model employs seven features: elapsed time, issue age, pricing interest rate, scenario interest rate, premium ratio, sum assured and mortality intensity. It predicts a standardized reserve ratio instead of raw reserve values, improving numerical stability across policies with different sums assured. On the test set the model achieved R² = 0.9887, MAE = 785.48 and RMSE = 1212.76. Inference on 200 policies was about 119.53 times faster than the classical solver. Results demonstrate strong predictive accuracy, physics consistency and robust boundary performance, while highlighting remaining challenges in monotonicity and out‑of‑distribution generalization.
Review: The platform effectively embeds physical constraints into deep learning, delivering fast and interpretable reserve predictions and offering a practical solution for large‑scale insurance scenario analysis.