引導式測驗 · 硬 等級

ADMET 與毒性預測測驗

解釋 ADMET 終點、QSAR 驗證、測定變異性和適用性領域,而不將預測視為安全性證明。

相關引導路徑Admet 與毒性預測
問題 1 的 8

What does ADMET stand for?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. What does ADMET stand for?
    • Absorption, distribution, metabolism, efficacy, and transport
    • Activity, dose, mechanism, efficacy, and testing
    • Absorption, distribution, metabolism, excretion, and toxicity
    • A single clinical safety score
  2. Why are ADMET predictions endpoint-specific?
    • Every assay measures the same safety result
    • Different assays measure distinct biological properties
    • Molecular fingerprints determine all outcomes directly
    • Toxicity can be inferred from solubility alone
  3. Why can a random molecular split overstate performance on new chemical scaffolds?
    • It prevents training examples from being used
    • It guarantees external validation
    • It removes assay noise
    • Related analogs may appear in both training and test sets
  4. What does an applicability domain describe?
    • The chemical and data conditions where the model has relevant training support
    • The set of molecules with the highest predicted safety scores
    • The dataset size used to train the model
    • A preselected safety cutoff for changing a probability into a class
  5. Why document assay source and protocol with training labels?
    • It guarantees every value is noise-free
    • It removes the need for external testing
    • Experimental methods and units can change what a measured value means
    • It determines molecule novelty
  6. What does a positive predicted safety result establish?
    • Proof that a compound is safe in humans
    • A model-based hypothesis that may guide testing
    • A replacement for toxicology assays
    • A guarantee across every dose
  7. Which split can better assess generalization to novel scaffolds?
    • A random split with close analogs in both sets
    • A training-only score
    • A split made after viewing test outcomes
    • A scaffold-separated split
  8. Which information supports transparent QSAR use?
    • Defined endpoint, model algorithm, applicability domain and validation evidence
    • Only a high training score
    • Only the molecule string
    • A single combined ADMET label