Cervical cancer is one of the most commonly appearing cancers, which early diagnosis is of greatest importance. Unfortunately, many diagnoses are based on subjective opinions of doctors—to date, there is no general measurement method with a calibrated standard. The problem can be solved with the measurement system being a fusion of an optoelectronic sensor and machine learning algorithm to provide reliable assistance for doctors in the early diagnosis stage of cervical cancer. We demonstrate the preliminary research on cervical cancer assessment utilizing an optical sensor and a prediction algorithm. Since each matter is characterized by refractive index, measuring its value and detecting changes give information about the state of the tissue. The optical measurements provided datasets for training and validating the analyzing software. We present data preprocessing, machine learning results utilizing four algorithms (Random Forest, eXtreme Gradient Boosting, Naïve Bayes, Convolutional Neural Networks) and assessment of their performance for classification of tissue as healthy or sick. Our solution allows for rapid sample measurement and automatic classification of the results constituting a potential support tool for doctors.

Cervical cancer remains a critical global health issue, with early detection and precise classification of cervical intraepithelialneoplasia (CIN) essential for effective management. This study introduces a machine learning-based approach to enhance theresolution of refractive index measurements, a key factor in diagnosing CIN and cervical cancer. We employed supervisedlearning to predict the refractive indices of liquid samples, conducting low-coherence interferometric measurements within therange of 1.3 to 1.5, with increments of 0.01. The optical spectra obtained were transformed into feature vectors tailored for aclassification model, which was then applied to a cervical tissue phantom made from liquids with known refractive indices. Thisallowed us to classify refractive index changes into five distinct categories: normal tissue, CIN-1, CIN-2, CIN-3, and cancerousstates. The model demonstrated high performance, with cross-validation scores showing a classification accuracy of 0.93,precision of 0.94, recall of 0.91, and an F1-score of 0.92 on the training/validation dataset. Testing results were similarly robust,with a classification accuracy of 0.91, precision of 0.90, recall of 0.93, and an F1-score of 0.91. These metrics confirm themodel’s high prognostic accuracy and its potential for application to new data. Additionally, our methodology enabled thegeneration of synthetic data mimicking experimental signals, extending the resolution of the acquired results. This enhancementoffers a promising tool for improving diagnostic methods based on refractive index measurements, potentially advancing earlydetection and classification in cervical cancer screening.