Artículo de revisión
Published on 20 de noviembre de 2025 | http://doi.org/10.5867/medwave.2025.10.3120
Artificial intelligence for skin lesion classification and diagnosis in dermatology: A narrative review
Back to article
Types of artificial intelligence [8].
| Deep learning | Uses deep neural networks to solve complex problems | Pattern recognition in images, natural language processing, and gaming | Requires substantial amounts of data, computationally intensive |
| Neural networks | Computational model inspired by the human brain, composed of interconnected nodes | Speech recognition, pattern recognition, image processing | Requires extensive training, complexity in interpretation |
| Reinforcement learning | An agent learns through interaction with an environment and receives feedback in the form of rewards or penalties | Gaming, robotics, and autonomous decision-making | Requires extended training time, design complexity |
| Feedforward neural networks | Information flows in one direction, without cycles or loops | Pattern recognition in images, natural language processing | Does not handle sequential data well, such as time series |
| Recurrent neural networks | Include cyclical connections to handle sequential data and retain information over time | Natural language processing, automatic translation, and time series | May face issues of vanishing or exploding gradients |
| Convolutional neural networks | Specifically designed for image processing, with shared weight layers and pooling layers | Object recognition in images, and medical diagnosis based on images | Requires enormous amounts of training data, specialized hardware |
| Decision trees | Tree-shaped models that make decisions based on features at each node | Medical diagnosis, customer classification, decision-making | May be prone to overfitting, difficulty capturing complex relationships |
| Random forests | An ensemble method that combines multiple decision trees to improve accuracy | Classification, regression, anomaly detection | Complexity in interpretation, resource-intensive |
| Support vector machines | Classifies data points by finding the hyperplane that best separates classes | Text classification, speech recognition, bioinformatics | May not perform well on large datasets, sensitive to the choice of kernel |
| Logistic regression | Linear model used for binary classification problems | Risk prediction, disease classification, and marketing | Limited to binary classification problems, does not handle nonlinear relationships |
Source: prepared by the authors based on the results of the study.