Models of logistic regression analysis, support vector machine, and back-propagation neural network based on serum tumor markers in colorectal cancer diagnosis

Machine Learning Analysis of Immune Cells for Diagnosis and Prognosis of Cutaneous Melanoma

Tumor infiltration, known to associate with various cancer initiations and progressions, is a promising therapeutic target for aggressive cutaneous melanoma. Then, the relative infiltration of 24 kinds of immune cells in melanoma was assessed by a single sample gene set enrichment analysis (ssGSEA) program from a public database. The multiple machine learning algorithms were applied to evaluate the efficiency of immune cells in diagnosing and predicting the prognosis of melanoma. In comparison with the expression of immune cell in tumor and normal control, we built the immune diagnostic models in training dataset, which can accurately classify melanoma patients from normal (LR AUC = 0.965, RF AUC = 0.99, SVM AUC = 0.963, LASSO AUC = 0.964, and NNET AUC = 0.989). These diagnostic models were also validated in three outside datasets and suggested over 90% AUC to distinguish melanomas from normal patients. Moreover, we also developed a robust immune cell biomarker that could estimate the prognosis of melanoma. This biomarker was also further validated in internal and external datasets. Following that, we created a nomogram with a composition of risk score and clinical parameters, which had high accuracies in predicting survival over three and five years. The nomogram's decision curve revealed a bigger net benefit than the tumor stage. Furthermore, a risk score system was used to categorize melanoma patients into high- and low-risk subgroups. The high-risk group has a significantly lower life expectancy than the low-risk subgroup. Finally, we observed that complement, epithelial-mesenchymal transition, and inflammatory response were significantly activated in the high-risk group. Therefore, the findings provide new insights for understanding the tumor infiltration relevant to clinical applications as a diagnostic or prognostic biomarker for melanoma.

Carbohydrate antigen 125, carbohydrate antigen 15-3 and low-density lipoprotein as risk factors for intraocular metastases in postmenopausal breast cancer

The prognosis of patients with postmenopausal breast cancer (PBC) could be improved by the early detection of intraocular metastases (IOMs). However, serum biomarkers for IOMs in PBC remain elusive. In the current study, we investigated patients with PBC, and compared serum parameters in an IOM and a non-IOM group, and then differentiated the risk factors related to IOMs. A comparison between an IOM and a non-IOM (NIOM) group was performed using Student t-test and a Chi-Squared test. After constructing a Poisson regression model to identify risk factors, we plotted receiver operating characteristic curves to evaluate the predictive value of significant risk factors in detecting IOMs. The incidence of IOMs in PBC was 1.16%. The histopathology results were not significantly different between the 2 groups. The levels of serum carbohydrate antigen 125 (CA-125), carbohydrate antigen 15-3 (CA15-3) and alkaline phosphatase were significantly elevated in IOMs compared with NIOMs (P = .082, P < .001, and P < .001, respectively). Compared with NIOMs, age, carbohydrate antigen 19 to 9, hemoglobin, calcium, total cholesterol, low-density lipoprotein (LDL) and apolipoprotein A1 were remarkably lower in IOMs (P = .038, P < .001, P < .001, P = .032, P = .041, P < .001, and P = .001, respectively). Poisson regression suggested that CA-125, CA15-3 and LDL were contributing to IOMs in PBC as risk factors (OR = 1.003, 95% CI: 1.001-1.005; OR = 1.025, 95% CI: 1.019-1.033; OR = 0.238, 95% CI: 0.112-0.505, respectively). A receiver operating characteristic curve revealed that the cut-off values for CA-125, CA15-3 and LDL were 16.78 0 U/mL, 63.175 U/mL, and 2.415 mmol/L, respectively. The combination of CA-125 and CA15-3 showed significant diagnostic value (area under the curve [AUC] = 0.982, P < .001). Our investigation suggests that CA-125, CA15-3 and LDL remarkably predict IOMs in PBC as risk factors, and the combination of CA-125 and CA15-3 shows considerable diagnostic value.

A Novel Method for Colorectal Cancer Screening Based on Circulating Tumor Cells and Machine Learning

Colorectal cancer is one of the most common types of cancer, and it can have a high mortality rate if left untreated or undiagnosed. The fact that CRC becomes symptomatic at advanced stages highlights the importance of early screening. The reference screening method for CRC is colonoscopy, an invasive, time-consuming procedure that requires sedation or anesthesia and is recommended from a certain age and above. The aim of this study was to build a machine learning classifier that can distinguish cancer from non-cancer samples. For this, circulating tumor cells were enumerated using flow cytometry. Their numbers were used as a training set for building an optimized SVM classifier that was subsequently used on a blind set. The SVM classifier’s accuracy on the blind samples was found to be 90.0%, sensitivity was 80.0%, specificity was 100.0%, precision was 100.0% and AUC was 0.98. Finally, in order to test the generalizability of our method, we also compared the performances of different classifiers developed by various machine learning models, using over-sampling datasets generated by the SMOTE algorithm. The results showed that SVM achieved the best performances according to the validation accuracy metric. Overall, our results demonstrate that CTCs enumerated by flow cytometry can provide significant information, which can be used in machine learning algorithms to successfully discriminate between healthy and colorectal cancer patients. The clinical significance of this method could be the development of a simple, fast, non-invasive cancer screening tool based on blood CTC enumeration by flow cytometry and machine learning algorithms.

Status quo and future prospects of artificial neural network from the perspective of gastroenterologists

Artificial neural networks (ANNs) are one of the primary types of artificial intelligence and have been rapidly developed and used in many fields. In recent years, there has been a sharp increase in research concerning ANNs in gastrointestinal (GI) diseases. This state-of-the-art technique exhibits excellent performance in diagnosis, prognostic prediction, and treatment. Competitions between ANNs and GI experts suggest that efficiency and accuracy might be compatible in virtue of technique advancements. However, the shortcomings of ANNs are not negligible and may induce alterations in many aspects of medical practice. In this review, we introduce basic knowledge about ANNs and summarize the current achievements of ANNs in GI diseases from the perspective of gastroenterologists. Existing limitations and future directions are also proposed to optimize ANN’s clinical potential. In consideration of barriers to interdisciplinary knowledge, sophisticated concepts are discussed using plain words and metaphors to make this review more easily understood by medical practitioners and the general public.

Towards Novel Non-Invasive Colorectal Cancer Screening Methods: A Comprehensive Review

Colorectal cancer (CRC) is one of the leading cancer-related causes of death in the world. Since the 70s, many countries have adopted different CRC screening programs, which has resulted in a decrease in mortality. However, current screening test options still present downsides. The commercialized stool-based tests present high false-positive rates and low sensitivity, which negatively affects the detection of early stage carcinogenesis. The gold standard colonoscopy has low uptake due to its invasiveness and the perception of discomfort and embarrassment that the procedure may bring. In this review, we collected and described the latest data about alternative CRC screening techniques that can overcome these disadvantages. Web of Science and PubMed were employed as search engines for studies reporting on CRC screening tests and future perspectives. The searches generated 555 articles, of which 93 titles were selected. Finally, a total of 50 studies, describing 14 different CRC alternative tests, were included. Among the investigated techniques, the main feature that could have an impact on CRC screening perception and uptake was the ease of sample collection. Urine, exhaled breath, and blood-based tests promise to achieve good diagnostic performance (sensitivity of 63-100%, 90-95%, and 47-97%, respectively) while minimizing stress and discomfort for the patient.

Developing a clinical decision support system based on the fuzzy logic and decision tree to predict colorectal cancer

Background: Colorectal Cancer (CRC) is the most prevalent digestive system- related cancer and has become one of the deadliest diseases worldwide. Given the poor prognosis of CRC, it is of great importance to make a more accurate prediction of this disease. Early CRC detection using computational technologies can significantly improve the overall survival possibility of patients. Hence this study was aimed to develop a fuzzy logic-based clinical decision support system (FL-based CDSS) for the detection of CRC patients. Methods: This study was conducted in 2020 using the data related to CRC and non-CRC patients, which included the 1162 cases in the Masoud internal clinic, Tehran, Iran. The chi-square method was used to determine the most important risk factors in predicting CRC. Furthermore, the C4.5 decision tree was used to extract the rules. Finally, the FL-based CDSS was designed in a MATLAB environment and its performance was evaluated by a confusion matrix. Results: Eleven features were selected as the most important factors. After fuzzification of the qualitative variables and evaluation of the decision support system (DSS) using the confusion matrix, the accuracy, specificity, and sensitivity of the system was yielded 0.96, 0.97, and 0.96, respectively. Conclusion: We concluded that developing the CDSS in this field can provide an earlier diagnosis of CRC, leading to a timely treatment, which could decrease the CRC mortality rate in the community.

Comparison of Diagnosis Accuracy between a Backpropagation Artificial Neural Network Model and Linear Regression in Digestive Disease Patients: an Empirical Research

Introduction

A Noninvasive diagnosis model for digestive diseases is the vital issue for the current clinical research. Our systematic review is aimed at demonstrating diagnosis accuracy between the BP-ANN algorithm and linear regression in digestive disease patients, including their activation function and data structure.

Methods

We reported the systematic review according to the PRISMA guidelines. We searched related articles from seven electronic scholarly databases for comparison of the diagnosis accuracy focusing on BP-ANN and linear regression. The characteristics, patient number, input/output marker, diagnosis accuracy, and results/conclusions related to comparison were extracted independently based on inclusion criteria.

Results

Nine articles met all the criteria and were enrolled in our review. Of those enrolled articles, the publishing year ranged from 1991 to 2017. The sample size ranged from 42 to 3222 digestive disease patients, and all of the patients showed comparable biomarkers between the BP-ANN algorithm and linear regression. According to our study, 8 literature demonstrated that the BP-ANN model is superior to linear regression in predicting the disease outcome based on AUROC results. One literature reported linear regression to be superior to BP-ANN for the early diagnosis of colorectal cancer.

Conclusion

The BP-ANN algorithm and linear regression both had high capacity in fitting the diagnostic model and BP-ANN displayed more prediction accuracy for the noninvasive diagnosis model of digestive diseases. We compared the activation functions and data structure between BP-ANN and linear regression for fitting the diagnosis model, and the data suggested that BP-ANN was a comprehensive recommendation algorithm.

The prognostic value of preoperative serum CA724 for CEA-normal colorectal cancer patients

Background and Purpose

There had been no recognized serum tumor marker to predict the prognosis of colorectal cancer (CRC) patients with normal preoperative serum carcinoembryonic antigen (CEA) levels. The purpose of this study was to determine whether preoperative serum carbohydrate antigen 724 (CA724) was of predictive function for the prognosis of CRC patients with normal CEA levels.

Methods

The medical records of 295 CRC patients with normal CEA levels who underwent surgery at the Department of Colorectal Anal Surgery of the First Affiliated Hospital of Guangxi Medical University (Guangxi, China) between September 2012 and September 2014 were retrospectively reviewed. The Chi-square test was used to test the correlation between preoperative serum CA724 levels and clinical features. Kaplan–Meier curves were conducted to calculate the overall survival (OS) rate and disease-free survival (DFS) of patients. Cox regression analysis was applied to conduct univariate and multivariate analysis of the following four preoperative serum tumor makers namely CA724, carbohydrate antigen 199 (CA199), carcinoembryonic antigen 125 (CA125), carcinoembryonic antigen 242 (CA242) and clinical features. Nomograms for prognostic parameter of OS and DFS were developed using R v3.2.5.

Results

In the Chi-square test, only pathological node stage (pN stage) (X² = 14.514, P = 0.001) and differentiation (X² = 10.712, P = 0.001) were associated with serum CA724 levels. In the Kaplan–Meier analysis, the results revealed that the OS and DFS in patients with high CA724 was poorer than those with normal. In the multivariate Cox regression analysis of OS and DFS, only pT stage, pN stage, metastasis and serum CA724 were independent prognostic risk factors for CRC patients with normal CEA levels.

Conclusion

Preoperative serum CA724 might serve as a potential prognostic factor for CRC patients with normal serum CEA levels.

Clinical characteristics of sentinel polyps and their correlation with proximal colon cancer: A retrospective observational study

BACKGROUND

Colorectal cancer is a common malignant tumor of the digestive tract. The relationship between sentinel polyps (rectal polyps with proximal colon cancer) and proximal colon cancer has received extensive attention in recent years. However, there is still no clear conclusion regarding the relationship.

AIM

To investigate the clinical characteristics of sentinel polyps and their correlation with proximal colon cancer.

METHODS

A retrospective analysis of 2587 patients with rectal polyps from January 2006 to December 2017 was performed. According to whether or not proximal colon cancer was diagnosed, the patients were divided into either a sentinel polyp group (192 patients) or a pure rectal polyp group (2395 patients). The endoscopic features, clinicopathological features, therapeutic effects, and short-term prognosis were analyzed and compared between the two groups.

RESULTS

The mean age of patients in the sentinel polyp group was generally higher than that of the pure rectal polyp group, and the positivity rates of anemia, stool occult blood, and tumor markers of the sentinel polyp group were also significantly higher than those in the rectal polyp group (χ² = 90.56, P < 0.01; χ² = 70.30, P < 0.01; χ² = 92.80, P < 0.01). The majority of the patients in the sentinel polyp group had multiple polyps, large polyps, adenomatous polyps, or sessile polyps (χ² = 195.96, P < 0.01; χ² = 460.46, P < 0.01; χ² = 94.69, P < 0.01; χ² = 48.01, P < 0.01). Most of the proximal colon cancers were Duke’s A and B stages in the sentinel polyp group. In the pure rectal polyp group, 2203 patients underwent endoscopic treatment, and all of the patients were cured and discharged. In the sentinel polyp group, 65 patients underwent radical operation, and 61 patients received endoscopic submucosal dissection or endoscopic mucosal resection. Additionally, 21 patients were lost to follow-up after 6-12 mo, and the loss rate was 10.94%. A total of 63.16% of patients experienced remission without tumor recurrence or metastasis, 33.33% of patients experienced tumors regression or improved symptoms, and the other 3.51% of the patients died.

CONCLUSION

If there are multiple, sessile, and adenomatous rectal polyps with a maximum diameter > 1 cm, the possibility of the carcinogenesis of the polyps or of the proximal colon should be monitored closely. These patients should be followed in the short-term and should undergo a whole-colon examination.

Artificial intelligence applications for pediatric oncology imaging

Machine learning algorithms can help to improve the accuracy and efficiency of cancer diagnosis, selection of personalized therapies and prediction of long-term outcomes. Artificial intelligence (AI) describes a subset of machine learning that can identify patterns in data and take actions to reach pre-set goals without specific programming. Machine learning tools can help to identify high-risk populations, prescribe personalized screening tests and enrich patient populations that are most likely to benefit from advanced imaging tests. AI algorithms can also help to plan personalized therapies and predict the impact of genomic variations on the sensitivity of normal and tumor tissue to chemotherapy or radiation therapy. The two main bottlenecks for successful AI applications in pediatric oncology imaging to date are the needs for large data sets and appropriate computer and memory power. With appropriate data entry and processing power, deep convolutional neural networks (CNNs) can process large amounts of imaging data, clinical data and medical literature in very short periods of time and thereby accelerate literature reviews, correct diagnoses and personalized treatments. This article provides a focused review of emerging AI applications that are relevant for the pediatric oncology imaging community.

Establishment and characterization of a novel cell line (cc‑006cpm8) of moderately/poorly differentiated colorectal adenocarcinoma derived from a primary tumor of a patient

In the present study, the cc‑006cpm8 novel colon cell line was established from a sample of right colorectal adenocarcinoma obtained from a woman with liver metastasis. It was possible to culture this cell line for ≥100 passages in vitro with vigorous growth. Morphologically, the cells grew as several layers with tight adhesion to the surface of the culture plate. The morphological, immunological and ultrastructural features of these cells suggested their epithelial origin. The characterization of this cell line indicated a doubling time of 27 h, a colony forming efficiency of 73.2% in semisolid media and a plate efficiency of 66.5% in liquid culture. The modal number of chromosomes was 50. In vivo, the cc‑006cpm8 cells underwent tumorigenesis in all nude mice used. Immunohistochemical analysis demonstrated that mutS homolog 2 (MSH2) and MSH6 were expressed; however, mutL homolog 1 and postmeiotic segregation 2 were downregulated in cc‑006cpm8 cells. To determine the mutation profile of the cell line analyzed, exome capture DNA sequencing was performed. The results revealed 20 hypermutated exons comprising single nucleotide polymorphisms, and insertion and deletions (InDels), including single nucleotide variants of mucin (MUC)19, MUC16, MUC12, filaggrin and AHNAK nucleoprotein 2, and InDels of β defensin‑126, microRNA‑3665, WNK lysine deficient protein kinase 1 and SLAIN motif‑containing protein 1. In addition, commonly mutated genes in colorectal cancer and exon mutations of genes in cc‑006cpm8 cells were analyzed, including adenomatous polyposis coli, tumor protein p53, Drosophila mothers against decapentaplegic 4, phosphatidylinositol‑4,5‑bisphosphate 3‑kinase catalytic subunit α and Kirsten rat sarcoma, and genes associated with the DNA mismatch repair pathway were investigated.

A reliable method for colorectal cancer prediction based on feature selection and support vector machine

Colorectal cancer (CRC) is a common cancer responsible for approximately 600,000 deaths per year worldwide. Thus, it is very important to find the related factors and detect the cancer accurately. However, timely and accurate prediction of the disease is challenging. In this study, we build an integrated model based on logistic regression (LR) and support vector machine (SVM) to classify the CRC into cancer and normal samples. From various factors, human location, age, gender, BMI, and cancer tumor type, tumor grade, and DNA, of the cancer, we select the most significant factors (p < 0.05) using logistic regression as main features, and with these features, a grid-search SVM model is designed using different kernel types (Linear, radial basis function (RBF), Sigmoid, and Polynomial). The result of the logistic regression indicates that the Firmicutes (AUC 0.918), Bacteroidetes (AUC 0.856), body mass index (BMI) (AUC 0.777), and age (AUC 0.710) and their combined factors (AUC 0.942) are effective for CRC detection. And the best kernel type is RBF, which achieves an accuracy of 90.1% when k = 5, and 91.2% when k = 10. This study provides a new method for colorectal cancer prediction based on independent risky factors. Graphical abstract Flow chart depicting the method adopted in the study. LR (logistic regression) and ROC curve are used to select independent features as input of SVM. SVM kernel selection aims to find the best kernel function for classification by comparing Linear, RBF, Sigmoid, and Polynomial kernel types of SVM, and the result shows the best kernel is RBF. Classification performance of LR + RF, LR + NB, LR + KNN, and LR + ANNs models are compared with LR + SVM. After these steps, the cancer and healthy individuals can be classified, and the best model is selected.

Diagnosis of Compressed Nerve Root in Lumbar Disc Herniation Patients by Surface Electromyography

OBJECTIVE

To establish a logistic regression model using surface electromyography (SEMG) parameters for diagnosing the compressed nerve root at L₅ or S₁ level in patients with lumbar disc herniation (LDH).

METHODS

This study recruited 24 patients with L₅ nerve root compression and 23 patients with S₁ nerve root compression caused by LDH from May 2014 to May 2016. SEMG signals from the bilateral tibialis anterior and lateral gastrocnemius were measured. The root mean square (RMS), the RMS peak time, the mean power frequency (MPF), and the median frequency (MF) were analyzed. The accuracy, sensitivity, and specificity values were calculated separately. The areas under the curve (AUC) of the receiver-operating characteristic (ROC) curve and the kappa value were used to evaluate the accuracy of the SEMG diagnostic model.

RESULTS

The accuracy of the SEMG model ranged from 85.71% to 100%, with an average of 93.57%. The sensitivity, specificity, AUC, and kappa value of the logistic regression model were 0.98 ± 0.05, 0.92 ± 0.09, 0.95 ± 0.04 (P = 0.006), and 0.87 ± 0.11, respectively (P = 0.001). The final diagnostic model was: P=1-11+ey; y = 10.76 - (5.95 × TA_RMS Ratio) - (0.38 × TA_RMS Peak Time Ratio) - (5.44 × 44 × LG_RMS Peak Time Ratio). L₅ nerve root compression is diagnosed when P < 0.5 and S₁ nerve root compression when P ≥ 0.5.

CONCLUSIONS

The logistic regression model developed in this study showed high diagnostic accuracy in detecting the compressed nerve root (L₅ and S₁ ) in these patients with LDH.

Blood-Based Protein Signatures for Early Detection of Colorectal Cancer: A Systematic Review

OBJECTIVES

Blood-based proteins might be an attractive option for early detection of colorectal cancer (CRC), but individually they are unlikely to achieve the diagnostic performance required for population based screening. We aimed at summarizing current evidence of diagnostic performance of signatures based on multiple proteins for early detection of CRC.

METHODS

A systematic literature review adhering to the PRISMA (preferred reporting items for systematic reviews and meta-analysis) guidelines was performed. PubMed and Web of Science databases were searched for potentially relevant studies published until 28th August, 2017. Relevant studies were identified by predefined eligibility criteria. Estimates of indicators of diagnostic performance such as sensitivity, specificity, and the area under the curve (AUC), along with information on validation and other key methodological procedures were extracted. Study quality was assessed by a QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies 2) instrument tool.

RESULTS

Thirty six eligible studies with numbers of CRC cases ranging from 23 to 512 and the number of proteins included in signatures ranged from 3 to 13 were identified. Reported Youden's Index and AUC ranged from 0.19 to 0.95 and from 0.62 to 0.996, respectively. However most studies, especially those reporting better diagnostic performance, were conducted in clinical rather than screening setting and many studies lacked any internal or external validation of identified algorithm.

CONCLUSIONS

Blood-based tests using signatures of multiple proteins may be a promising approach for non-invasive CRC screening. However, promising signatures identified in clinical settings still require rigorous evaluation in large studies conducted in true screening setting.

International scientific communications in the field of colorectal tumour markers

AIM

To analyze scientometrically the dynamic science internationalization on colorectal tumour markers as reflected in five information portals and to outline the significant journals, scientists and institutions.

METHODS

A retrospective problem-oriented search was performed in Web of Science Core Collection (WoS), MEDLINE, BIOSIS Citation Index (BIOSIS) and Scopus for 1986-2015 as well as in Dervent Innovations Index (Derwent) for 1995-2015. Several specific scientometric parameters of the publication output and citation activity were comparatively analyzed. The following scientometric parameters were analyzed: (1) annual dynamics of publications; (2) scientific institutions; (3) journals; (4) authors; (5) scientific forums; (6) patents - number of patents, names and countries of inventors, and (7) citations (number of citations to publications by single authors received in WoS, BIOSIS Citation Index and Scopus).

RESULTS

There is a trend towards increasing publication output on colorectal tumour markers worldwide along with high citation rates. Authors from 70 countries have published their research results in journals and conference proceedings in 21 languages. There is considerable country stratification similar to that in most systematic investigations. The information provided to end users and scientometricians varies between these data-bases in terms of most parameters due to different journal coverage, indexing systems and editorial policy. The lists of the so-called “core” journals and most productive authors in WoS, BIOSIS, MEDLINE and Scopus along with the list of the most productive authors - inventors in Derwent present a particular interest to the beginners in the field, the institutional and national science managers and the journal editorial board members. The role of the purposeful assessment of scientific forums and patents is emphasized.

CONCLUSION

Our results along with this problem-oriented collection containing the researchers’ names, addresses and publications could contribute to a more effective international collaboration of the coloproctologists from smaller countries and thus improve their visibility on the world information market.