Variability of echo state network prediction horizon for partially observed dynamical systems.

Study of dynamical systems using partial state observation is an important problem due to its applicability to many real-world systems. We address the problem by studying an echo state network (ESN) framework with partial state input with partial or full state output. Application to the Lorenz system and Chua's oscillator (both numerically simulated and experimental systems) demonstrate the effectiveness of our method. We show that the ESN, as an autonomous dynamical system, is capable of making short-term predictions up to a few Lyapunov times. However, the prediction horizon has high variability depending on the initial condition-an aspect that we explore in detail using the distribution of the prediction horizon. Further, using a variety of statistical metrics to compare the long-term dynamics of the ESN predictions with numerically simulated or experimental dynamics and observed similar results, we show that the ESN can effectively learn the system's dynamics even when trained with noisy numerical or experimental data sets. Thus, we demonstrate the potential of ESNs to serve as cheap surrogate models for simulating the dynamics of systems where complete observations are unavailable.

Population Pharmacokinetics and Exposure-Response with Teclistamab in Patients With Relapsed/Refractory Multiple Myeloma: Results From MajesTEC-1.

BACKGROUND: Teclistamab, a B-cell maturation antigen × CD3 bispecific antibody, is approved in patients with relapsed/refractory multiple myeloma (RRMM) who have previously received an immunomodulatory agent, a proteasome inhibitor, and an anti-CD38 antibody. OBJECTIVE: We report the population pharmacokinetics of teclistamab administered intravenously and subcutaneously (SC) and exposure-response relationships from the phase I/II, first-in-human, open-label, multicenter MajesTEC-1 study. METHODS: Phase I of MajesTEC-1 consisted of dose escalation and expansion at the recommended phase II dose (RP2D; 1.5 mg/kg SC weekly, preceded by step-up doses of 0.06 and 0.3 mg/kg); phase II investigated the efficacy of teclistamab RP2D in patients with RRMM. Population pharmacokinetics and the impact of covariates on teclistamab systemic exposure were assessed using a 2-compartment model with first-order absorption for SC and parallel time-independent and time-dependent elimination pathways. Exposure-response analyses were conducted, including overall response rate (ORR), duration of response (DoR), progression-free survival (PFS), overall survival (OS), and the incidence of grade ≥ 3 anemia, neutropenia, lymphopenia, leukopenia, thrombocytopenia, and infection. RESULTS: In total, 4840 measurable serum concentration samples from 338 pharmacokinetics-evaluable patients who received teclistamab were analyzed. The typical population value of time-independent and time-dependent clearance were 0.449 L/day and 0.547 L/day, respectively. The time-dependent clearance decreased rapidly to < 10% after 8 weeks of teclistamab treatment. Patients who discontinue teclistamab after the 13th dose are expected to have a 50% reduction from Cmax in teclistamab concentration at a median (5th to 95th percentile) time of 15 days (7-33 days) after Tmax and a 97% reduction from Cmax in teclistamab concentration at a median time of 69 days (32-163 days) after Tmax. Body weight, multiple myeloma type (immunoglobulin G vs non-immunoglobulin G), and International Staging System (ISS) stage (II vs I and III vs I) were statistically significant covariates on teclistamab pharmacokinetics; however, these covariates had no clinically relevant effect on the efficacy of teclistamab at the RP2D. Across all doses, ORR approached a plateau at the concentration range associated with RP2D, and in patients who received the RP2D, a flat exposure-response curve was observed. No apparent relationship was observed between DoR, PFS, OS, and the incidence of grade ≥3 adverse events across the predicted exposure quartiles. CONCLUSION: Body weight, myeloma type, and ISS stage impacted systemic teclistamab exposure without any clinically relevant effect on efficacy. The exposure-response analyses for ORR showed a positive trend with increasing teclistamab systemic exposure, with a plateau at the RP2D, and there was no apparent exposure-response trend for safety or other efficacy endpoints. These analyses support the RP2D of teclistamab in patients with RRMM. CLINICAL TRIAL REGISTRATION: NCT03145181 (phase I, 09 May 2017); NCT04557098 (phase II, 21 September 2020).

Effect of discrete time observations on synchronization in Chua model and applications to data assimilation.

Data assimilation is a tool, which incorporates observations in the model to improve the forecast, and it can be thought of as a synchronization of the model with observations. This paper discusses results of numerical identical twin experiments, with observations acting as master system coupled unidirectionally to the slave system at discrete time instances. We study the effects of varying the coupling constant, the observational frequency, and the observational noise intensity on synchronization and prediction in a low dimensional chaotic system, namely, the Chua circuit model. We observe synchrony in a finite range of coupling constant when coupling the x and y variables of the Chua model, but not when coupling the z variable. This range of coupling constant decreases with increasing levels of noise in the observations. The Chua system does not show synchrony when the time gap between observations is greater than about one-seventh of the Lyapunov time. Finally, we also note that the prediction errors are much larger when noisy observations are used than when using observations without noise.