Phase 1: Preparatory Work

Step 1: Defining Research Goals and Boundaries

Action: Formulate the main research question in writing

Method: Use the "5W+H" technique (What? Who? Where? When? Why? How?)

Example: Instead of "Artificial intelligence in medicine" → "How are deep CNN-based computer vision methods applied for the diagnosis of malignant skin neoplasms in 2023-2025?"

Result: Clearly formulated research question with specific boundaries

Step 2: Researching Key Terminology

Action: Compile a list of main terms, concepts, and synonyms

Method: Conduct preliminary mini-research in open sources

Example:

• Main terms: deep convolutional neural networks, CNN, dermatoscopy
• Synonyms: computer vision = machine vision
• Specific concepts: melanoma, ISIC dataset, ROC AUC and Dice metrics

Result: Structured dictionary of terms for maximally precise query

Step 3: Determining Relevance Criteria

Action: Establish parameters for evaluating information quality

Method: Define timeframes, source authority, types of evidence

Example:

• Timeframes: 2023-2025
• Authoritative sources: peer-reviewed journals, leading conferences (CVPR, ICCV, MICCAI)