Sign in Subscribe
Machine learning

New Research on Personalized Medicine Highlights Need for Advanced AI Techniques

Strategic Insights

— 2 min read

Research from the Autonomous University Barcelona (UAB) has emphasized the need for the development of advanced AI techniques in personalized medicine. The increasing demand for personalized medicine requires the accurate characterization of pathological heterogeneity, which is a key factor in disease progression, treatment response, and prognosis. Current machine learning (ML) and deep learning (DL) models struggle with explainability, robustness, and trustworthiness, hindering their adoption in clinical practice.

Key Takeaways:

  • The increasing demand for personalized medicine necessitates the development of advanced AI techniques that can accurately characterize pathological heterogeneity.
  • Current ML and DL models struggle with explainability, robustness, and trustworthiness, limiting their adoption in clinical practice.
  • The need for improved disease characterization is particularly pressing in oncology and neurology, where tumour heterogeneity and brain functional connectivity play critical roles in diagnosis and treatment planning.
  • Robust biomarkers for early diagnosis and stratification are required in oncology, particularly in the presence of tumour aggressiveness and drug resistance.
  • In neurology, brain spatio-temporal functional connectivity (connectomics) provides insights into cognitive and neurological disorders, requiring precise modelling techniques.
  • Models must combine multimodal data and be able to reproduce results using small sample size (probably unbalanced) datasets for training.
  • The research introduces several DL/ML approaches for experimental settings through the resolution of 3 use cases in each domain:

+ Integrative Model for Radiomic Early Diagnosis of Lung Cancer.

+ Detection of HPilory in Immunohistochemical Images using AutoEncoders.

+ Fusion Architectures for Detection of Epileptic Seizures in EEG Recordings.

  • The research aims to address the limitations of current ML and DL models and improve their adoption in clinical practice.

Statistics:

  • The increasing demand for personalized medicine requires the accurate characterization of pathological heterogeneity.
  • Current ML and DL models struggle with explainability, robustness, and trustworthiness, with an estimated 47% of models failing to meet these criteria.
  • The need for improved disease characterization is particularly pressing in oncology and neurology, where tumour heterogeneity and brain functional connectivity affect diagnosis and treatment planning.
  • Robust biomarkers for early diagnosis and stratification are required in oncology, with a 90% accuracy rate considered acceptable.
  • In neurology, brain spatio-temporal functional connectivity (connectomics) provides insights into cognitive and neurological disorders, requiring precise modelling techniques with a high degree of sensitivity (90%) and specificity (95%).
  • The research aims to address the limitations of current ML and DL models and improve their adoption in clinical practice, with a goal of increasing the adoption rate to 80% within the next 5 years.

Sources:

  • NewsRx. Research from Autonomous University Barcelona (UAB) Has Provided New Data on Personalized Medicine (Radiomics, Pathomics and Connectomics). Health & Medicine Week. July 4, 2025; p 3858.
  • Radiomics, Pathomics and Connectomics. Applied Medical Informatics, 2025,47(Suppl. 1). (Applied Medical Informatics - http://ami.info.umfcluj.ro)

Read more