Ignorar datos no medibles en modelos matemáticos de tumores puede distorsionar resultados

Researchers have demonstrated that predictive mathematical models in oncology yield misleading results if they ignore unmeasurable data. When analyzing tumor measurements, it was found that including only available data generated significant biases, overestimating initial growth and underestimating later phases. This finding underscores the critical need to incorporate all information, even that which escapes direct detection, to avoid erroneous conclusions about treatment efficacy. This study is fundamental to the field of open data because it exposes how selection bias in public or shared datasets can distort reproducible science. When researchers publish only successful measurements, they omit vital structural information about detection limits. Ignoring these “censored” data not only compromises individual accuracy but also degrades the aggregate quality of open repositories, hindering the development of robust, universal models that depend on comprehensive transparency of scientific information. The proposed solution is a mathematical framework that estimates the probability of unmeasurable tumors, thereby improving the precise identification of parameters. This implies that researchers must abandon the practice of discarding incomplete information and design protocols that capture richer data. By sharing both measurable and unmeasurable data, the scientific community ensures that data-driven models based on open data reflect biological reality with greater fidelity, facilitating genuine advances in medicine and other disciplines sensitive to data quality.

Source: larepublica.es
Published on 2025-03-21

Ignorar datos no medibles en modelos matemáticos de tumores puede distorsionar resultados

Related news