Introduction: As the world's population is ageing, the prevalence of cognitive dysfunction and dementia are increasing. A better understanding of the drivers of cognitive decline is imperative in order to help identify potential targets for preventative strategies. Several risk factors are associated with an increased risk of developing MCI and dementia, including, female sex, age, educational attainment, lifestyle, and dietary habits[1]. In addition, genetic factors are thought to play an important role in the development of dementia, particularly Alzheimer’s disease, and may interact with various lifestyle factors. Genome-wide association studies (GWAS) enable the detection of risk factors associated with cognitive dysfunction[2], while machine learning techniques can facilitate a comprehensive analysis of complex datasets[3] enabling the study of interactions of genetics with a range of other factors, nutritional status, lifestyle, and environmental, in ageing populations. The aim of this study is to employ GWAS coupled with machine learning techniques to identify genetic markers associated with cognitive dysfunction in older adults, while exploring potential interactions with lifestyle factors.
Methods: The TUDA cohort, comprising 5186 older adults aged 60 to 102 recruited across the Island of Ireland, provides comprehensive data on nutrition, health, lifestyle factors, and more recently, genotyping of single nucleotide polymorphisms (SNPs). The SNP data underwent rigorous quality control measures at both the SNP and individual levels. A GWAS was performed using logistic regression to identify SNPs associated with cognitive dysfunction as assessed using the Repeatable Battery for the Assessment of Neuropsychological Assessment (RBANS), an age-adjusted and sensitive neuropsychiatric test. All SNPs significantly associated with cognitive dysfunction (RBANS score <70) were then integrated with detailed sociodemographic, clinical, biochemical, nutritional, and lifestyle data to train 3 classifiers - decision trees, Naïve Bayes, and random forests - to identify key predictors of cognitive performance.
Results: Preliminary analysis revealed two SNPs associated with cognitive dysfunction from the Manhattan plot; SNPs rs3771791 (P = 8.41 x 10-8) and rs1056815951 (P = 1.02 x 10-7) after adjusting for multiple testing by Bonferroni correction, but did not reach genome-wide significance. Further investigation is warranted to validate these findings and elucidate their functional significance in relation to cognitive dysfunction.
Conclusion: In conclusion, we identified two significant SNPs associated with cognitive dysfunction, underscoring the potential of genetic markers in predicting cognitive decline. This preliminary analysis highlights how the integration of GWAS and machine learning techniques can shed light on the multifaceted determinants of cognitive dysfunction in ageing populations. Further research is needed to validate these findings and explore their functional implications, paving the way for personalised interventions aimed at mitigating age-related cognitive decline.
References:
1. Campbell NL, Unverzagt F, LaMantia MA et al (2013) Clin Geriatr Med 29, 873-893.
2. Bertram L and Tanzi RE (2009) Hum Mol Genet 18, 37-145.
3. Kirk D, Kok E, Tufano M et al (2022) Adv Nutr 13, 2573-2589.

**Introduction:** As the world's population is ageing, the prevalence of cognitive dysfunction and dementia are increasing. A better understanding of the drivers of cognitive decline is imperative in order to help identify potential targets for preventative strategies. Several risk factors are associated with an increased risk of developing MCI and dementia, including, female sex, age, educational attainment, lifestyle, and dietary habits[1]. In addition, genetic factors are thought to play an important role in the development of dementia, particularly Alzheimer’s disease, and may interact with various lifestyle factors. Genome-wide association studies (GWAS) enable the detection of risk factors associated with cognitive dysfunction[2], while machine learning techniques can facilitate a comprehensive analysis of complex datasets[3] enabling the study of interactions of genetics with a range of other factors, nutritional status, lifestyle, and environmental, in ageing populations. The aim of this study is to employ GWAS coupled with machine learning techniques to identify genetic markers associated with cognitive dysfunction in older adults, while exploring potential interactions with lifestyle factors.
**Methods:** The TUDA cohort, comprising 5186 older adults aged 60 to 102 recruited across the Island of Ireland, provides comprehensive data on nutrition, health, lifestyle factors, and more recently, genotyping of single nucleotide polymorphisms (SNPs). The SNP data underwent rigorous quality control measures at both the SNP and individual levels. A GWAS was performed using logistic regression to identify SNPs associated with cognitive dysfunction as assessed using the Repeatable Battery for the Assessment of Neuropsychological Assessment (RBANS), an age-adjusted and sensitive neuropsychiatric test. All SNPs significantly associated with cognitive dysfunction (RBANS score <70) were then integrated with detailed sociodemographic, clinical, biochemical, nutritional, and lifestyle data to train 3 classifiers - decision trees, Naïve Bayes, and random forests - to identify key predictors of cognitive performance.
**Results:** Preliminary analysis revealed two SNPs associated with cognitive dysfunction from the Manhattan plot; SNPs rs3771791 (P = 8.41 x 10-8) and rs1056815951 (P = 1.02 x 10-7) after adjusting for multiple testing by Bonferroni correction, but did not reach genome-wide significance. Further investigation is warranted to validate these findings and elucidate their functional significance in relation to cognitive dysfunction.
**Conclusion:** In conclusion, we identified two significant SNPs associated with cognitive dysfunction, underscoring the potential of genetic markers in predicting cognitive decline. This preliminary analysis highlights how the integration of GWAS and machine learning techniques can shed light on the multifaceted determinants of cognitive dysfunction in ageing populations. Further research is needed to validate these findings and explore their functional implications, paving the way for personalised interventions aimed at mitigating age-related cognitive decline.
**References:**
1. Campbell NL, Unverzagt F, LaMantia MA _et al_ (2013) _Clin Geriatr Med_ 29, 873-893.
2. Bertram L and Tanzi RE (2009) _Hum Mol Genet_ 18, 37-145.
3. Kirk D, Kok E, Tufano M _et al_ (2022) _Adv Nutr_ 13, 2573-2589.

LB7 - Exploration of genetic factors in relation to the ageing brain: preliminary genomic analysis with machine learning approaches from the TUDA study.

Unique to the new Congress (previously the Summer Conference) is that you will have greater opportunity to engage with more nutrition science than ever before. In 2024 the core programme will be delivered by the Irish Section of The Nutrition Society, supported by the Science Committee. The theme of these symposia will be ‘new data – focused approaches and challenges’. Outside of the core programme you will have the chance to hear the latest research from the Society’s partner organisations, universities, the private sector and more, on a diverse range of research.