Traditional drug development has long relied on hypothesis-driven approaches. This method involves pinpointing specific targets and mechanisms through preclinical studies, but often grapples with uncertainties regarding the translatability of these findings to real-world patient settings. Enter contemporary research and development, which is now benefiting from a human-first, hypothesis-agnostic approach. By leveraging extensive datasets from large human populations that include comprehensive biomarker and health information, researchers are identifying disease drivers more accurately and developing targeted, effective medicines supported by causal human evidence.
A groundbreaking innovation in this realm is MILTON (MachIne Learning with phenoType associatONs), a new machine learning-based tool recently introduced in Nature Genetics. Developed by the Centre for Genomics Research, MILTON enhances case-control association statistics from large cohort studies. This AI tool’s strength lies in its ability to reclassify individuals who might have been inaccurately categorized in these studies, thus significantly boosting the statistical power of genetic association analyses.
While larger sample sizes are crucial for genetic discoveries, MILTON provides a novel approach to maximize the value of existing samples. It excels at extracting meaningful signals from random noise, potentially broadening the scope and accuracy of gene discovery for numerous diseases.
MILTON’s capabilities extend beyond improving case-control studies; it also excels in predicting individuals at risk for future disease diagnoses. This dual functionality enhances its role in early disease detection and preventive healthcare. Early intervention is critical for many complex diseases, which are often diagnosed only after clinical symptoms emerge. MILTON’s integration of molecular profiles could enable earlier and more precise detection, paving the way for tailored preventative or early intervention therapies.
Developed using a robust dataset of 67 routine clinical biomarkers from nearly 500,000 participants in the UK Biobank, MILTON demonstrates remarkable predictive power. It integrates multi-omic data, including measurements from 3,000 plasma proteins, to enhance prediction accuracy for numerous diseases. When benchmarked against traditional polygenic risk scores, MILTON showcased superior predictive capabilities for most studied conditions.
As we continue to advance the boundaries of genomic research, MILTON represents a significant step forward in transforming drug discovery and preventative healthcare. By harnessing the power of machine learning and comprehensive datasets, we are not only improving our understanding of diseases but also refining our approach to early intervention and treatment.
Article written by iSanidad
15/09/2024
Source:
iSanidad
https://isanidad.com/293140/la-nueva-ia-de-astrazeneca-milton-promete-predecir-mas-de-mil-enfermedades-antes-del-diagnostico/