Table 2.
Comparative analysis of AI applications across omics fields in lung cancer research.
| Omics Field | AI Techniques | Key Applications | Performance/Strengths | Limitations |
|---|---|---|---|---|
| Genomics | Deep Learning, Machine Learning | - Gene mutation analysis | - High accuracy in mutation detection | - Requires large datasets |
| - Genomic instability assessment | - Improved early diagnosis | - Complexity in interpreting genetic variations | ||
| - Epigenetics study | - Non-invasive screening (liquid biopsy) | |||
| - Liquid biopsy enhancement | ||||
| Transcriptomics | Deep Learning, Natural Language Processing | - Gene expression analysis | - Identification of novel biomarkers | - Challenges in handling noise in expression data |
| - RNA sequencing data interpretation | - Insight into gene regulation | - Difficulty in interpreting complex gene interactions | ||
| Proteomics | Machine Learning, Deep Neural Networks | - Protein-protein interaction prediction | - High-throughput analysis | - Limited by proteome complexity |
| - Biomarker discovery | - Identification of drug targets | - Challenges in data integration | ||
| Metabolomics | Machine Learning, Deep Learning | - Metabolic pathway analysis | - Early diagnosis potential | - Metabolite variability challenges |
| - Biomarker identification | - Treatment response monitoring | - Need for standardized data collection | ||
| Immunomics | Deep Learning, Machine Learning | - Immune escape mechanism study | - Personalized immunotherapy approaches | - Complexity of immune system interactions |
| - Neoantigen discovery | - Improved understanding of tumor-immune interactions | - Limited availability of comprehensive immune data | ||
| - Immunotherapy response prediction | ||||
| Microbiomics | Machine Learning | - Microbiome profile analysis | - Novel insights into lung cancer etiology | - Challenges in standardizing microbiome data |
| - Host-microbiome interaction study | - Potential for microbiome-based therapies | - Complexity in interpreting microbial diversity | ||
| Radiomics | Convolutional Neural Networks, Deep Learning | - Image feature extraction | - Non-invasive diagnosis | - Variability in imaging protocols |
| - Tumor classification | - High accuracy in image analysis | - Need for large, diverse image datasets | ||
| - Treatment response prediction | ||||
| Pathomics | Convolutional Neural Networks, Deep Learning | - Automated tissue analysis | - Improved diagnostic accuracy | - Challenges in standardizing tissue preparation |
| - Cancer subtype classification | - Efficient histopathological analysis | - Need for extensive pathologist validation |