A reinforcement learning-based hybrid modeling framework for bioprocess kinetics identification.

Constructing predictive models to simulate complex bioprocess dynamics, particularly time-varying (i.e., parameters varying over time) and history-dependent (i.e., current kinetics dependent on historical culture conditions) behavior, has been a longstanding research challenge. Current advances in hybrid modeling offer a solution to this by integrating kinetic models with data-driven techniques. This article proposes a novel two-step framework: first (i) speculate and combine several possible kinetic model structures sourced from process and phenomenological knowledge, then (ii) identify the most likely kinetic model structure and its parameter values using model-free Reinforcement Learning (RL). Specifically, Step 1 collates feasible history-dependent model structures, then Step 2 uses RL to simultaneously identify the correct model structure and the time-varying parameter trajectories. To demonstrate the performance of this framework, a range of in-silico case studies were carried out. The results show that the proposed framework can efficiently construct high-fidelity models to quantify both time-varying and history-dependent kinetic behaviors while minimizing the risks of over-parametrization and over-fitting. Finally, the primary advantages of the proposed framework and its limitation were thoroughly discussed in comparison to other existing hybrid modeling and model structure identification techniques, highlighting the potential of this framework for general bioprocess modeling.

Whether individualized dose escalation should be recommended for lymph nodes with different sizes in the definitive radiotherapy of cervical cancer?

BACKGROUND AND PURPOSE: Dose escalation for positive node maybe improve the regional control of patients with node-positive cervical cancer, but the optimal dose for nodes of different sizes remains controversial. The purpose of this study was to explore the individualized dose escalation for lymph nodes (LNs) with different sizes in the definitive radiotherapy of cervical cancer. METHODS: A total of 1002 cervical cancer patients with the International Federation of Gynecology and Obstetrics (FIGO 2009) stage IB1-IVA, who were treated by definitively radiotherapy between September 2013 and December 2016 were enrolled. All LNs identified by computed tomography/magnetic resonance imaging (CT/MRI) were assigned into three groups according to the short diameters of < 1 cm, 1-2 cm or ≥ 2 cm at pretreatment. RESULTS: In total, 580 patients with 1310 LNs were detected. The nodal control rate in groups of LNs < 1 cm, 1-2 cm and ≥ 2 cm was 99.4%, 96%, and 75.9%, respectively (P = 0.000). Among LNs < 1 cm, the control, overall survival (OS) and progression-free survival (PFS) rates did not significantly differ among three dose-based groups (≤ 50.4 Gy, 50.4-60 Gy, > 60 Gy) (control rate, 99.4% vs. 99.3% vs. 100%, P = 0.647) (5-year OS, 76.2% vs. 79% vs. 81.6%, P = 0.682) (5-year PFS, 74.1% vs. 73.9% vs. 78.9% P = 0.713). Among LNs of 1-2 cm, the control and PFS rates were significantly higher in the group of dose ≥ 55 Gy than the group of dose < 55 Gy (control rate, 98% vs. 93.6%, P = 0.028) (5-year PFS, 69.6% vs. 56.7%, P = 0.025). However, this did not cause a significant difference for 5-year OS rate (72.6% vs. 68.3%, P = 0.5). Among LNs ≥ 2 cm, the control, OS, and PFS rates were higher in the group of dose ≥ 55 Gy than the group of dose < 55 Gy, while no significant difference was found (control rate, 82.1% vs. 63.2%, P = 0.107) (5-year OS, 60.6% vs. 37.5%, P = 0.141) (5-year PFS, 51.5% vs.37.5%, P = 0.232). CONCLUSIONS: Radiation dose escalation is not necessary for LNs < 1 cm, and dose escalation of 55 Gy is enough for LNs of 1-2 cm.