Development of predictive machine learning models based on salivary multiomic data for oral cance...

Development of predictive machine learning models based on salivary multiomic data for oral cancer diagnosis

Oral cancer represents a major public health challenge in Spain, with an incidence of 6.7 cases per 100,000 inhabitants and an annual mortality rate of over 1,500 deaths, especially in men over 50 years of age. Five-year survival is between 50% and 60%, improving significantly in cases detected in early stages. However, diagnosis is often made in advanced stages, which compromises prognosis and increases healthcare costs.

Oral squamous cell carcinoma accounts for 90% of cases and is closely related to tobacco and alcohol consumption. Currently, biopsy is the gold standard diagnostic method, but its invasiveness and limited follow-up make it difficult to use routinely.

In this context, clinical salivaomics emerges as a promising alternative within precision medicine. This discipline comprehensively analyses the molecular components of saliva - including genome, proteome, metabolome and microbiome - using advanced omics technologies. Its non-invasiveness, accessibility and stability make saliva an ideal fluid for early diagnosis, follow-up and monitoring of oral cancer.

Objectives

This project aims to develop predictive models based on the analysis of the microbiome and salivary proteome - using 16S rRNA sequencing and SWATH proteomics technology - applying machine learning techniques for the diagnosis of oral cancer. Accurate and generalisable multi-omics tools will be designed to classify disease, estimate individual risk, validate performance with external data and consider key clinical variables such as age, gender and toxic habits.