Mortality Prediction as Boundary Value Problem

Zhorin, Victor (2026): Mortality Prediction as Boundary Value Problem.

Abstract

We present a mathematical framework for mortality prediction from discrete medical event sequences, formulating the problem as a boundary value problem on causal path space. Discrete event sequences lift to rough path space; signatures provide coordinate-free trajectory analysis. The transformer architecture emerges as computing weighted projections of path signatures, with Time2Vec temporal encoding providing spectral parameterization of the rough path lift. We establish connections to tropical geometry: ReLU networks compute tropical rational functions, and the decision boundary is a tropical hypersurface interpretable as the viscosity solution to a Hamilton-Jacobi equation on trajectory space. Mortality acts as absorbing boundary condition, with gradients propagating retroactively to reshape early event representations - the computational realization of ”boundary conditions determine interior.” The framework unifies perspectives from causal set theory (discrete causal ordering as primitive), rough path theory (coordinate-free trajectory invariants), tropical geometry (piecewise-linear structure from max-plus algebra), and viscosity solutions (weak solutions selected by vanishing diffusion). The architecture’s parameter efficiency is explained by alignment with the problem’s intrinsic mathematical structure: tropical decision boundaries, Lie algebraic dimension reduction, and binary boundary factorization.

• Item Type: MPRA Paper
• Language: English
• Keywords: Hamilton-Jacobi-Bellman equation; viscosity solutions; heterogeneous agents; mortality prediction; transformer architecture; tropical geometry; rough path signatures; boundary value problem; mean field games; parameter efficiency
• Subjects: C - Mathematical and Quantitative Methods > C1 - Econometric and Statistical Methods and Methodology: General > C14 - Semiparametric and Nonparametric Methods: General; C - Mathematical and Quantitative Methods > C4 - Econometric and Statistical Methods: Special Topics > C45 - Neural Networks and Related Topics; C - Mathematical and Quantitative Methods > C6 - Mathematical Methods ; Programming Models ; Mathematical and Simulation Modeling > C61 - Optimization Techniques ; Programming Models ; Dynamic Analysis; C - Mathematical and Quantitative Methods > C6 - Mathematical Methods ; Programming Models ; Mathematical and Simulation Modeling > C63 - Computational Techniques ; Simulation Modeling; D - Microeconomics > D3 - Distribution > D31 - Personal Income, Wealth, and Their Distributions; G - Financial Economics > G2 - Financial Institutions and Services > G22 - Insurance ; Insurance Companies ; Actuarial Studies; I - Health, Education, and Welfare > I1 - Health > I14 - Health and Inequality

References

Achdou, Han, Lasry, Lions, Moll (2022). "Income and Wealth Distribution in Macroeconomics: A Continuous-Time Approach." Review of Economic Studies 89(1), 45–86.

Kaplan, Moll, Violante (2018). "Monetary Policy According to HANK." American Economic Review 108(3), 697–743.

Lasry, Lions (2007). "Mean Field Games." Japanese Journal of Mathematics 2(1), 229–260.

Moll, Ryzhik (2026). "Mean Field Games without Rational Expectations." Communications in Contemporary Mathematics [In Press].

Moll, Townsend, Zhorin (2017). "Economic Development, Flow of Funds, and the Equilibrium Interaction of Financial Frictions." PNAS 114(24), 6176–6184. (Also NBER WP 19618.)

Townsend (1994). "Risk and Insurance in Village India." Econometrica 62(3), 539–591.

Crandall, Lions (1983). "Viscosity Solutions of Hamilton-Jacobi Equations." Transactions of the AMS 277(1), 1–42.

Crandall, Ishii, Lions (1992). "User's Guide to Viscosity Solutions of Second Order Partial Differential Equations." Bulletin of the AMS 27(1), 1–67.

Jensen (1988). "The Maximum Principle for Viscosity Solutions of Fully Nonlinear Second Order Partial Differential Equations." Archive for Rational Mechanics and Analysis 101(1), 1–27.

Fleming, Soner (2006). Controlled Markov Processes and Viscosity Solutions. 2nd ed., Springer.

Maclagan, Sturmfels (2015). Introduction to Tropical Geometry. Graduate Studies in Mathematics 161, AMS.

Zhang, Naitzat, Zhong (2018). "Tropical Geometry of Deep Neural Networks." Proceedings of ICML 80, 5824–5832.

Litvinov (2007). "The Maslov Dequantization, Idempotent and Tropical Mathematics." Journal of Mathematical Sciences 140(3), 349–386.

Lyons (1998). "Differential Equations Driven by Rough Signals." Revista Matemática Iberoamericana 14(2), 215–310.

Chen (1958). "Integration of Paths — A Faithful Representation of Paths by Noncommutative Formal Power Series." Transactions of the AMS 89(2), 395–407.

Hambly, Lyons (2010). "Uniqueness for the Signature of a Path of Bounded Variation and the Reduced Path Group." Annals of Mathematics 171(1), 109–167.

Salvi, Cass, Foster, Sherlock, Sherlock (2021). "The Signature Kernel Is the Solution of a Goursat PDE." SIAM Journal on Mathematics of Data Science 3(3), 873–899.

Kidger, Lyons (2021). "Neural Controlled Differential Equations for Irregular Time Series." NeurIPS 2020.

Bombelli, Lee, Meyer, Sorkin (1987). "Space-Time as a Causal Set." Physical Review Letters 59(5), 521–524.

Sorkin (2003). "Causal Sets: Discrete Gravity." In Lectures on Quantum Gravity, Springer.

Hoel, Albantakis, Tononi (2013). "Quantifying Causal Emergence Shows That Macro Can Beat Micro." PNAS 110(49), 19790–19795.

Vaswani et al. (2017). "Attention Is All You Need." NeurIPS 2017.

Kazemi, Goel, Jain, Kobyzev, Sethi, Forsyth, Poupart (2019). "Time2Vec: Learning a General Purpose Representation of Time." arXiv:1907.05321.

Cox (1972). "Regression Models and Life-Tables." Journal of the Royal Statistical Society B 34(2), 187–220.

Sawhney, Blacker, Sigursson, Bhatt (2023). "Life2Vec: Predicting Life Outcomes Using Sequences of Life Events." Nature Computational Science 4, 43–56.