Page 64 - BH-3-3

Brain & Heart AI in biomarker discovery for CVDs

identifies potential targets for therapeutic intervention. This As AI continues to evolve, its integration into the
integrative approach is especially effective in handling multi- field of biomarker discovery for CVDs will become
omics data – genomics, proteomics, and metabolomics – more sophisticated and impactful. This advancement
facilitating the discovery of novel biomarkers and offering a underscores the need for ongoing research and
more comprehensive view of disease mechanisms. Real-time collaboration across technological and medical
monitoring of CVD patients through wearable technology communities. In addition, future studies should focus
is gaining traction, with AI’s ability to continuously analyze on the interactions between specific circulating miRNAs
data from these devices leading to timely interventions and and protein biomarkers to identify indicators capable
improved disease management. This ongoing feedback of predicting the presence, progression, and treatment
enhances both patient and healthcare provider insights into responses in cancer and other diseases. These studies
cardiovascular health. typically involve high-throughput sequencing and
proteomics, supported by extensive bioinformatics and
Despite these advancements, implementing AI in
healthcare faces significant challenges, such as concerns statistical analyses to establish significant correlations. As
AI technology advances, its application in the discovery
about data privacy and the need for standardized protocols and validation of biomarkers for CVDs becomes
for effective integration of AI tools into existing systems. increasingly indispensable.
Ensuring that AI models are interpretable is vital for
building trust among healthcare professionals. Table 2 illustrates specific ways in which AI technologies
can enhance various aspects of biomarker research, from
Clinicians and scientists are increasingly incorporating
AI into their workflows, where it standardizes, accelerates, standardizing procedures to predictive modeling of disease
progression based on biomarker data. Each application
and expands the scope of traditional processes. not only contributes to refining existing methods but
14
Innovations like AI-driven clinical decision support also opens new avenues for personalized medicine in
systems are revolutionizing liquid biopsy applications cardiovascular care.
by thoroughly analyzing vast patient data sets, including
circulating free DNA profiles, treatment histories, and 4. Perspectives and challenges
clinical outcomes. These systems facilitate the creation
of predictive models and detailed treatment protocols. As omics technologies and AI continue to evolve, the
16
15
Notably, the recent reviews by Khera et al. and Sun et al. landscape of biomarker CVDs is undergoing a significant
highlight the potential of AI to redefine cardiovascular transformation. Below are several key developments that
practice and research. They discuss how AI-driven could reshape cardiovascular healthcare:
innovations in diagnostic modalities and digital-native (i) Enhanced diagnostic accuracy. AI algorithms are adept
biomarkers of disease are setting new frontiers in at analyzing complex datasets, including genetic
cardiovascular care, offering expanded access to screening information, imaging data, and clinical records. This
and monitoring, particularly for underserved populations, capability allows for the identification of nuanced
and propelling biological and clinical discoveries toward biomarkers, leading to more accurate and earlier
more personalized, precise, and effective care. diagnoses of CVDs. For instance, AI could detect
subtle patterns in echocardiograms that precede
In a recent study by DeGroat et al., a novel integration visible symptoms of heart disease.
17
of AI and traditional statistical methods was used to identify (ii) Risk stratification. By leveraging digital biomarkers
biomarkers for CVDs. The research involved analyzing from diverse data sources, AI can stratify patients
the complete transcriptome of CVD patients, initially based on their risk levels. This facilitates personalized
using Pearson correlation, Chi-square tests, and analysis treatment approaches and more effective management
of variance to compare transcriptomic and clinical data of individuals with cardiovascular conditions,
against healthy controls. Subsequently, a recursive feature potentially adjusting therapies based on predicted risk
elimination classifier ranked relevant transcriptomic shifts.
features, with the top 10% undergoing further evaluation (iii) Predictive analytics. AI models excel at predicting
through four ML classifiers: Random Forest, Support the progression of diseases by analyzing trends and
Vector Machine, Xtreme Gradient Boosting Decision Trees, patterns within patient data. Such analytics enable
and k-Nearest Neighbors. After optimizing the models proactive interventions that can mitigate or even
with a soft voting classifier, the approach achieved up to prevent severe health outcomes, especially in patients
96% accuracy in distinguishing CVD patients from healthy identified as high-risk.
individuals, identifying 18 significant transcriptomic (iv) Drug development and clinical trials. The acceleration
biomarkers that might enhance early CVD detection. in discovering new biomarkers through AI not only

Volume 3 Issue 3 (2025) 4 doi: 10.36922/bh.8442