Kovlyagina I, Jaric I. Phasing out animal research prematurely will maintain gender inequities in medicine. Nature Neuroscience 29, 1269–1270 (2026).
https://doi.org/10.1038/s41593-026-02309-w
In their comment paper, Irina Kovlyagina and Ivana Jaric argue that prematurely phasing out animal research could perpetuate existing inequalities in women’s health. Decades of reliance on male animals have left a considerable gap in our understanding of female physiology, disease mechanisms and treatment responses. Thus, ending in vivo research before these gaps are addressed would preserve a male-biased evidence base, particularly in areas involving development, pregnancy, systemic regulation, chronic stress and pharmacokinetics, which current non-animal methods cannot yet adequately reproduce. Furthermore, they stress that in vitro and artificial intelligence methods can perpetuate the same sex biases present in the data and biological materials on which they are based. Hence, while not defending the status quo of animal research, the authors stress that rapidly phasing it out is not a neutral act, and call for decisions on which animal studies should be phased out - and when to do so - to account for sex-related knowledge gaps.
Bellucci A, Baranowski BJ & Wright DC. The effects of housing temperature on mouse physiology and implications for disease modeling. Lab Animal 55, 108–116 (2026).
https://doi.org/10.1038/s41684-026-01696-8
This review discusses how housing mice below their thermoneutral zone, as occurs under conventional laboratory conditions, affects their physiology and may confound experimental results. Cold exposure produces behavioural and physiological adaptations, including increased energy expenditure and altered immune function, with implications for models of metabolic disorders, inflammation, cancer and exercise physiology. Housing temperature can also alter responses to pharmacological interventions through its effects on energy expenditure and body mass. This warrants considering environmental effects on the physiology of laboratory mice, and stresses the importance of reporting housing details, particularly room temperature, for cross-study comparison and reproducibility of results. Moreover, as several physiological responses to chronic cold are sex-specific, the authors consider it critical to include both sexes in study design.
Christine Ro. My research on mice is causing me stress. How can I become more resilient? Nature. Career Feature |14 July 2026.
https://doi.org/10.1038/d41586-026-01128-0
This Nature Career Feature brings attention to the stress and compassion fatigue experienced by some researchers, particularly early-career ones, in conducting procedures on laboratory animals. Christine Ro interviewed three scientists on this subject, who suggested finding mentoring elsewhere when supervisors fail to acknowledge legitimate concerns, searching for peer support, receiving training to improve animal welfare and refinement of procedures or even changing research direction. Also, proper self-care was deemed essential, such as getting enough rest, sleep, exercise, and meaningful connections with other people, particularly through harder times. Importantly, researchers should remind themselves of the importance of their work, while acknowledging the lives of the animals killed, rather than trying to desensitize themselves, as empathy, in the words of Fernando Gonzalez-Uarquin, is “a precursor of responsible science and integrity”.
Baran SW and Gaburro S. Hybrid mechanistic–machine learning PK/PD models with digital biomarkers: from cage to clinic. Frontiers in Pharmacology 17:1815118 (2026).
https://doi.org/10.3389/fphar.2026.1815118
This review examines how mechanistic pharmacokinetic and pharmacodynamic models can be combined with machine learning and continuous digital biomarkers. Mechanistic models provide a biologically interpretable framework, while machine learning can identify complex patterns, covariates and residual relationships that animal models may fail to capture. Digital data from home-cage monitoring and telemetry in animals, as well as from wearable devices and remote monitoring in patients, could moreover provide comparable physiological and behavioural outcomes across preclinical and clinical research. The authors describe potential improvements in parameter estimation, individual dose prediction and cross-species scaling, while stressing that greater computational complexity does not necessarily produce better models. They highlight overfitting, data leakage, device drift, missing data and poor generalisability as particular risks, and call for rigorous validation, transparency, reproducibility and a clearly defined context, rather than algorithmic novelty.