Liquid biopsy-based approaches offer many new opportunities to measure molecular biomarkers for the diagnosis, prognosis, and monitoring of disease. Artificial Intelligence (AI), such as machine learning, and its ability to identify signatures of specific disease states in multiplexed data, will be key to taking advantage of the new molecular information that microchip-based diagnostics can extract. Focusing on the liquid biopsy and applications of AI in medicine over years, Creative Biolabs is confident in providing high-quality AI-based liquid biopsy services to global researchers to promote disease research.

Introduction of AI in Liquid Biopsy

Researchers in the field of liquid biopsy are developing technologies to measure sparse molecular biomarkers shed from inaccessible tissue in easily sampled bodily fluids, such as urine, blood, saliva, sweat, feces, and tears. The last decade has seen great progress in this field, and there is a growing list of circulating indicators rare: circulating cells, microvesicles, nucleic acids, proteins, and metabolites that can be detected. Machine learning encompasses a set of computational techniques widely applied in many fields to reduce large numbers of measurements into lower-dimensional outputs that are more useful. In recent years, machine learning algorithms have been increasingly applied to liquid biopsy data to aid in disease diagnosis, prediction, and medical decision-making. A growing set of studies use these approaches to identify signatures in multiple circulating biomarkers for a wide range of applications, including cancer, tuberculosis, dengue fever, heart disease, liver disease, brain disease, and diabetes. These studies employ a variety of machine learning algorithms, including support vector machines, decision trees, and random forests, that outperform the sensitivity and specificity of individual markers in many applications.

A generic workflow for developing and evaluating a machine learning-based liquid biopsy diagnostic.Fig.1 A generic workflow for developing and evaluating a machine learning-based liquid biopsy diagnostic. (Ko, 2018)

Application of AI in Liquid Biopsy

Machine learning algorithms build a model from sample inputs and use that model to make predictions based on subsequent data. Generally, machine learning algorithms fall into two main categories: supervised and unsupervised learning. Supervised and unsupervised learning techniques are each useful for specific applications within the field of liquid biopsy.

  • Supervised learning

In supervised learning, the algorithm is provided a set of training data wherein the true state of the data is known, such as which subjects have cancer and which subjects are healthy. Based on this training data, the algorithm generates a model that is deployed to predict the state of subsequent subjects for which the true state is not known. These predictions can take the form of a 'classification problem,' identifying a set of discrete states (such as the stage of a patient's cancer), or a 'regression problem,' across a set of continuous variables (such as the volume of a developing tumor). Often, biomarkers are collected to characterize a group of subjects such that each subject has a single label (for example, which patients have or do not have the target condition). In these cases, algorithm training is limited by data collection and sample size, making supervised learning techniques most useful.

  • Unsupervised learning

In unsupervised learning, algorithms search for patterns in sets of data without labeled states. These algorithms, such as clustering methods, may be used to investigate the structure or distribution of a dataset, discover groups of similar examples within the data, or reduce data dimensionality. In the application of identifying tumor cells, unsupervised techniques (using a generative mixture model) have been used with performance approaching that of a supervised method (support vector machine) without the need for labeled data. Similar unsupervised approaches have also been used to identify circulating tumor cells based on genetic clustering, identify cell-free DNA of tumor origin, and classify tumor-educated platelets based on RNA expression profiles.

Services at Creative Biolabs

AI and Liquid Biopsy for Disease Research

Focusing on the development of in vitro diagnostics (IVD) over years, Creative Biolabs has thrown our sight into the AI-based liquid biopsy and done a lot of research on the subjects. With strong foundations and rich experience, we are confident in providing customer-satisfied AI-based liquid biopsy development services to global clients.

As an industry-leading biotechnology company, Creative Biolabs aspires to bring every client high-quality services and top-rated customer experiences. If you are interested in AI-based liquid biopsy development services for disease research or you have any other questions about our services, please don't hesitate to contact us for more information.


  1. Ko, J.; et al. Machine learning to detect signatures of disease in liquid biopsies - a user's guide. Lab Chip. 2018, 18(3): 395-405.

For Research Use Only.

Online Inquiry

*E-mail Address:
*Service & Products Interested:
Project Description:

Contact Us




Follow us on:
Copyright © 2021 Creative Biolabs.
Inquiry Basket