Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

QCS-ADME: Quantum Circuit Search for Drug Property Prediction with Imbalanced Data and Regression Adaptation

Kangyu Zheng, Tianfan Fu, Zhiding Liang

TL;DR

This work addresses ADME property prediction with quantum circuit search by adapting the Elivagar scoring framework to handle imbalanced classification and regression. It introduces a class-imbalance weight in RepCap and a Gaussian-reference for regression, enabling a training-free score to correlate with test outcomes via $Score(C) = CNR(C)^{\alpha_{CNR}} \times RepCap(C)$ and $RepCap(C)=1 - \frac{\|R_c - R_w \otimes R_{ref}\|^2_2}{2 \cdot n_c \cdot d_c^2}$, while for regression, $R_{ref} = \exp(-\frac{\|y_i - y_j\|_2^2}{2\sigma^2})$ guides state similarity. Experiments on PAMPA (classification) and Caco2 (regression) show that the revised scoring correlates moderately with performance in classification and can yield competitive MSE in regression (best 0.0955) relative to classical baselines, though ranking accuracy for regression remains challenging. The study demonstrates the potential of QCS in drug-property prediction and outlines future work toward hardware validation and broader dataset coverage.

Abstract

The biomedical field is beginning to explore the use of quantum machine learning (QML) for tasks traditionally handled by classical machine learning, especially in predicting ADME (absorption, distribution, metabolism, and excretion) properties, which are essential in drug evaluation. However, ADME tasks pose unique challenges for existing quantum computing systems (QCS) frameworks, as they involve both classification with unbalanced dataset and regression problems. These dual requirements make it necessary to adapt and refine current QCS frameworks to effectively address the complexities of ADME predictions. We propose a novel training-free scoring mechanism to evaluate QML circuit performance on imbalanced classification and regression tasks. Our mechanism demonstrates significant correlation between scoring metrics and test performance on imbalanced classification tasks. Additionally, we develop methods to quantify continuous similarity relationships between quantum states, enabling performance prediction for regression tasks. This represents the first comprehensive approach to searching and evaluating QCS circuits specifically for regression applications. Validation on representative ADME tasks-one imbalanced classification and one regression-demonstrates moderate positive correlation between our scoring metrics and circuit performance, significantly outperforming baseline scoring methods that show negligible correlation.

QCS-ADME: Quantum Circuit Search for Drug Property Prediction with Imbalanced Data and Regression Adaptation

TL;DR

This work addresses ADME property prediction with quantum circuit search by adapting the Elivagar scoring framework to handle imbalanced classification and regression. It introduces a class-imbalance weight in RepCap and a Gaussian-reference for regression, enabling a training-free score to correlate with test outcomes via and , while for regression, guides state similarity. Experiments on PAMPA (classification) and Caco2 (regression) show that the revised scoring correlates moderately with performance in classification and can yield competitive MSE in regression (best 0.0955) relative to classical baselines, though ranking accuracy for regression remains challenging. The study demonstrates the potential of QCS in drug-property prediction and outlines future work toward hardware validation and broader dataset coverage.

Abstract

The biomedical field is beginning to explore the use of quantum machine learning (QML) for tasks traditionally handled by classical machine learning, especially in predicting ADME (absorption, distribution, metabolism, and excretion) properties, which are essential in drug evaluation. However, ADME tasks pose unique challenges for existing quantum computing systems (QCS) frameworks, as they involve both classification with unbalanced dataset and regression problems. These dual requirements make it necessary to adapt and refine current QCS frameworks to effectively address the complexities of ADME predictions. We propose a novel training-free scoring mechanism to evaluate QML circuit performance on imbalanced classification and regression tasks. Our mechanism demonstrates significant correlation between scoring metrics and test performance on imbalanced classification tasks. Additionally, we develop methods to quantify continuous similarity relationships between quantum states, enabling performance prediction for regression tasks. This represents the first comprehensive approach to searching and evaluating QCS circuits specifically for regression applications. Validation on representative ADME tasks-one imbalanced classification and one regression-demonstrates moderate positive correlation between our scoring metrics and circuit performance, significantly outperforming baseline scoring methods that show negligible correlation.

Paper Structure

This paper contains 12 sections, 4 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Overview of Elivagar Scoring
  4. Handling Class Imbalance
  5. Regression Task Adaptation
  6. Experiments
  7. Dataset Selection and Preprocessing
  8. Training and Evaluation Configuration
  9. Results
  10. Performance on Imbalanced Classification
  11. Regression Task Performance
  12. Conclusion

Figures (3)

  • Figure 1: Overview of QCS-ADME
  • Figure 2: Unbalance classification results. (a): Spearman correlation between original final score and test loss of Elivagar 10.1145/3620665.3640354. (b): Spearman correlation between revised final score and test loss.
  • Figure 3: Regression results. (a): Spearman correlation between revised score and test loss of Elivagar 10.1145/3620665.3640354. (b): Spearman correlation between ranking relation of predicted and ground truth and revised score.