Search

RBF neural networks are well established tools for classification and regression problems. This article adapts a hybrid genetic algorithm to estimate the main parameters of the network. The proposed method utilizes a genetic algorithm in a conjunction with a local search procedure and a termination rule. The method is tested against other RBF variants on a series of well-known problems from the relevant literature and the results are reported.

A novel hybrid method is proposed for neural network training. The method consists of two phases: in the first phase the bounds for the neural network parameters are estimated using a genetic algorithm that uses intervals as chromosomes. In the second phase a genetic algorithm is used to train the neural network inside the bounding box located by the first phase. The proposed method is tested on a series of well-known datasets from the relevant literature and the results are reported.

Recently, there has been a growing research interest in utilizing the electroencephalogram (EEG) as a non-invasive diagnostic tool for neurodegenerative diseases. This article provides a detailed description of a resting-state EEG dataset of individuals with Alzheimer’s disease and frontotemporal dementia, and healthy controls. The dataset was collected using a clinical EEG system with 19 scalp electrodes while participants were in a resting state with their eyes closed. The data collection process included rigorous quality control measures to ensure data accuracy and consistency. The dataset contains recordings of 36 Alzheimer’s patients, 23 frontotemporal dementia patients, and 29 healthy age-matched subjects. For each subject, the Mini-Mental State Examination score is reported. A monopolar montage was used to collect the signals. A raw and preprocessed EEG is included in the standard BIDS format. For the preprocessed signals, established methods such as artifact subspace reconstruction and an independent component analysis have been employed for denoising. The dataset has significant reuse potential since Alzheimer’s EEG Machine Learning studies are increasing in popularity and there is a lack of publicly available EEG datasets. The resting-state EEG data can be used to explore alterations in brain activity and connectivity in these conditions, and to develop new diagnostic and treatment approaches. Additionally, the dataset can be used to compare EEG characteristics between different types of dementia, which could provide insights into the underlying mechanisms of these conditions.

Touch is a fundamental aspect of human interaction with the surrounding environment. It affects individuals' development in different manners and figures prominently in everyday operations such as the sense of presence, object recognition, performing actions, non-verbal communication and emotional state. In recent years there has been a growth of interest in researching the electro physiological activity of the brain originating from haptic stimulation. In the present preliminary experiment, we performed a classification process of extracted EEG features acquired from four healthy participants' EEG data when they actively touched different natural textures. Each participant was asked to use their fingertips and calmly rub for one minute, each of the three different textured materials (smooth, rough and water surface). EEG recordings were acquired and processed. Next, time and frequency-based features were extracted and used as inputto four classifiers to correctly identify each different texture. The results obtained show a classification performance of 63% with C4.5 algorithm and 76% with Random Forests and 10-fold cross-validation.

Non-Alcoholic Fatty Pancreas Disease (NAFPD) is the most common pancreatic condition in adults and is usually associated with obesity and insulin resistance. It is a new medical term that indicates the development of pancreatic steatosis, which at an advanced stage leads to the irreversible replacement of acinar cells with fat droplets. Although increasing prevalence rates are recorded worldwide for this condition, it has been studied to a small extent due to the diagnostic limitations of noninvasive medical imaging methods. In recent years and with the development of modern computer vision systems, digital pathology through biopsy imaging systems has become the gold standard in modern clinical trials. The current work presents an automated diagnostic tool for measuring the fat ratio in pancreatic biopsy specimens. The automated analysis is performed on a set of 20 histological images using supervised machine learning algorithms. Its diagnostic performance presents a minimum fat quantification error of 0.23% compared to that obtained from human semi-quantitative estimates.

In any hospital health care monitoring system, it is necessary to constantly monitor the patient's physiological parameters. Previous studies report that important parameters of any patient that have to be monitored in hospital are heart rate, respiratory rate, oxygen saturation, temperature, change in systolic blood pressure, motion, posture and its location. This work presents a monitoring system that has the capability to monitor in real-time the physiological parameters of the patient using a comfortable wearable device. A Wireless Sensor Network in collaboration with multiple wireless relay nodes, are responsible for collecting and sending the signals from the wireless sensors to the base station. The data is stored and processed using intelligent techniques in a cloud-based environment. Early warning alerts are automatically sent to the medical staff allowing them to intervene on time and earlier than when using manual vital sign observations. By grouping patients according to risk, medical experts can continuously monitor patients health status and quickly identify those that need their attention most.

A new approach for the assessment of motor symptoms caused by Parkinson's disease (PD), based on data recorded using a smart mobile phone, is presented in this manuscript. Data were obtained from the online platform kaggle.com, and were analyzed based on machine learning techniques to produce a comprehensive physician report, presenting motor symptoms in everyday life. The idea of this study is to equip PD patients with a smartphone, which will monitor signals in real time for a period of time in order to enhance the medical diagnosis routine of a physician, providing information about the overall picture of the patient's motor condition, resulting in the provision of individualized treatment to the patient.

This article introduces a method that can automatically classify the hemiplegia type (right or left side of the body is paralyzed) between healthy and non-healthy subjects. The proposed method utilizes the data taken from the accelerometer sensor of the RehaGait mobile gait analysis system. These data undergo a pre-processing and feature extraction stage before being sent as input to a scaled conjugate gradient backpropagation (SCG-BP) trained neural network. The proposed system is tested using a custom-created dataset containing 10 healthy and 20 patients suffering from hemiplegia (right or left). The experimental part of the system utilized 7 sensors placed on the left and right foot, the left and right shank, the left and right thigh, and the hip of each subject. Each sensor captured a 3-dimensional (3D) signal from 3 different device types: accelerometer, magnetometer, and gyroscope. The system utilized and split into 2-second windows only the accelerometer data, achieving a classification accuracy of 87.71%.

The prolongation of a wireless sensor network lifetime is closely related to the energy consumption of the most energy-consuming node of the network. As shown here, the energy consumption of a network is closely related to the nodes’ traffic load. In this sense, for minimizing the energy consumption of the most energy-consuming node and prolong the network's lifetime, a traffic load-based algorithm is proposed here. The proposed algorithm, by exploiting the results of a simple shortest path approach, discovers, for every node, the neighbors that hold the same distance (in hops) from the sink node, with the initially assigned parent. If such neighbors are found then all of them are used as parents and the traffic load is forwarded to all of them interchangeably. The proposed algorithm is also evaluated under simulation results, showing that it achieves the goals set and that it prolongs the network lifetime.

The paper introduces two modifications for the well-known PSO method to solve global optimization problems. The first modification deals with the termination of the method and the second with the bounding of the so-called velocity in order to prevent the method from creating particles outside the domain range of the objective function. The modified method was tested on a series of global optimization problems from the relevant literature and the results are reported.

The therapeutic efficacy of medication and treatment strategies is based on the diagnosis and staging of liver diseases. According to recent researches, the Collagen Proportional Area (CPA) is a reliable metric to assess fibrosis in liver tissues. Several image processing techniques are used for the analysis of liver biopsy images, providing objective assessment for the severity of the disease. In current work a novel classification via clustering algorithm is proposed, based on K-means, which is used for image segmentation of liver biopsies. More specifically, supervised learning is employed to insert constraints on centroids movement. Furthermore, feature weighting is utilized for the classification process. At first, a hypercube is extracted and feature weights are computed, for each class, using a training set of liver biopsy images. Classification via clustering follows, initializing a centroid for each class within the respective hypercube, which, during the iterations of the clustering, is allowed to move only inside the hypercube. The weighted Euclidean distance is used as similarity criterion to the clusters. 93 liver biopsy images are employed to evaluate the proposed approach. The classification results along with CPA values are computed.

Aim: Sex differences have long been reported in schizophrenia leading to the hypothesis that sex hormones may be implicated in the pathophysiology of the disorder. We assessed gonadal hormones during the fasted state in drug-naive patients with psychosis.Method: Fasting serum concentrations of follicular-stimulating hormone (FSH) and luteinizing hormone (LH), testosterone, free-testosterone, Sex Hormone Binding Globulin (SHBG) and oestradiol (E2) were compared between a group of 55 newly diagnosed, drug-naive, first-episode men with psychosis and a group of 55 healthy controls, matched for age, smoking status and BMI. Testosterone, free-testosterone and SHBG were compared between a group of 32 drug-naive, first-episode females with psychosis and a group of 32 healthy controls matched for age, smoking status and BMI.Results: Testosterone and free-testosterone levels were significantly lower in the patients' group and SHBG levels significantly higher in the patients' group compared to those in healthy controls. The two female groups had similar values in the hormones which were measured.Conclusion: Our findings provide evidence of lower testosterone and free-testosterone levels and increased SHBG levels in drug-naive, first-episode males with psychosis.KEY POINTSReduced testosterone and free-testosterone levels in drug-naive, first-episode males with psychosis.Increased SHBG levels in drug-naive first-episode males with psychosis.No difference in FSH, LH and E2 levels between drug-naive first episode males with psychosis and controls.No difference in testosterone, free-testosterone and SHBG levels between drug-naive, first-episode women with psychosis and controls.

In this paper, a hybrid learning approach, which combines the extreme learning machine (ELM) with a genetic algorithm (GA), is proposed. The utilization of this hybrid algorithm enables the creation of heterogeneous single layer neural networks (SLNNs) with better generalization ability than traditional ELM in terms of lower mean square error (MSE) for regression problems or higher accuracy for classification problems. The architecture of this method is not limited to traditional linear neurons, where each input participates equally to the neuron’s activation, but is extended to support higher order neurons which affect the network’s generalization ability. Initially, the proposed heterogeneous hybrid extreme learning machine (He-HyELM) algorithm creates a number of custom created neurons with different structure, which are used for the creation of homogeneous SLNNs. These networks are trained with ELM and an application specific GA evolves them into heterogeneous networks according to a fitness criterion utilizing the uniform crossover operator for the recombination process. After the completion of the evolution process, the network with the best fitness is selected as the most optimal. Experimental results demonstrate that the proposed learning algorithm can get better results than traditional ELM, homogeneous hybrid extreme learning machine (Ho-HyELM) and optimally pruned extreme learning machine (OP-ELM) for homogeneous and heterogeneous SLNNs.

Alzheimer's Disease (AD) is the most common type of dementia with world prevalence of more than 46 million people. The Mini-Mental State Examination (MMSE) score is used to categorize the severity and evaluate the disease progress. The electroencephalogram (EEG) is a cost-effective diagnostic tool and lately, new methods have developed for MMSE score correlation with EEG markers. In this paper, EEG recordings acquired from 14 patients with mild and moderate AD and 10 control subjects are analyzed in the five EEG rhythms (δ, θ, α, β, γ). Then, 38 linear and non-linear features are calculated. Multiregression linear analysis showed highly correlation of with MMSE score variation with Permutation Entropy of δ rhythm, Sample Entropy of θ rhythm and Relative θ power. Also, the best statistically significant regression models in terms of R2 are at O2 (0.542) and F4 (0.513) electrodes and at posterior (0.365) and left-temporal cluster (0.360).

"Dyslexia is one of the most frequent specific learning disorders which has often been associated with deficits in phonological awareness mainly caused by auditory and visual inabilities to recognize and discriminate phonemes and graphemes within words. Neuroimaging techniques like EEG recordings have been widely used to assess hemispheric differences in brain activation between students with dyslexia and their typical counterparts. Although dyslexia is a lifelong disorder which persists into adulthood, very few studies have been carried out targeting in adult population. In this study, we examined the brain activation differences between 14 typical (control group) and 12 university students with dyslexia (experimental group). The participants underwent two tasks consisting of 50 3-word groups characterized by different degrees of auditory and visual distinctiveness. The whole procedure was recorded with a 14-sensor sophisticated wearable EEG recording device (Emotiv EPOC+). The findings from the auditory task revealed statistically significant differences among the two sets of groups in the left temporal lobe in ?, ? and ? rhythms, in the left occipital lobe in ? rhythm, and in the right prefrontal area in ?, ? and ? rhythms, respectively. The students with dyslexia reported higher mean scores only in ? rhythm in the left temporal lobe, and in ?, ? and ? rhythms in the right prefrontal area. Concerning the visual task, statistically significant differences were evident in the left temporal lobe in ?, ? rhythms, in the occipital lobe in ?, ? and ? rhythms, in the parietal lobe in ? rhythm, and in the right occipital lobe in ?, ? and ? rhythms. The students with dyslexia reported higher mean scores only in the ? rhythm of both the left and right occipital lobe. The results indicate that there are differences in the hemispheric brain activation of students with or without dyslexia in various rhythms in both experimental conditions, thus, shedding light in the neurophysiological discrepancies between the two groups. It also lays great emphasis on the necessity of carrying out more studies in adult population with dyslexia."

In this paper we investigate the essential minimum functionality of the autonomous blockchain, and the minimum hardware and software required to support it in the micro-scale in the IoT world. The application of deep-blockchain operation in the lower-level activity of the IoT ecosystem, is expected to bring profound clarity and constitutes a unique challenge. Setting up and operating bit-level blockchain mechanisms on minimal IoT elements like smart switches and active sensors, mandates pushing blockchain engineering to the limits. “How deep can blockchain actually go?” “Which is the minimum Thing of the IoT world that can actually deliver autonomous blockchain functionality?” To answer, an experiment based on IoT micro-controllers was set. The “Witness Protocol” was defined to set the minimum essential micro-blockchain functionality. The protocol was developed and installed on a peer, ad-hoc, autonomous network of casual, real-life IoT micro-devices. The setup was tested, benchmarked, and evaluated in terms of computational needs, efficiency, and collective resistance against malicious attacks. The leading considerations are highlighted, and the results of the experiment are presented. Findings are intriguing and prove that fully autonomous, private micro-blockchain networks are absolutely feasible in the smart dust world, utilizing the capacities of the existing low-end IoT devices.

Alzheimer’s Disease (AD) is a neurodegenerative disorder and the most common type of dementia with a great prevalence in western countries. The diagnosis of AD and its progression is performed through a variety of clinical procedures including neuropsychological and physical examination, Electroencephalographic (EEG) recording, brain imaging and blood analysis. During the last decades, analysis of the electrophysiological dynamics in AD patients has gained great research interest, as an alternative and cost-effective approach. This paper summarizes recent publications focusing on (a) AD detection and (b) the correlation of quantitative EEG features with AD progression, as it is estimated by Mini Mental State Examination (MMSE) score. A total of 49 experimental studies published from 2009 until 2020, which apply machine learning algorithms on resting state EEG recordings from AD patients, are reviewed. Results of each experimental study are presented and compared. The majority of the studies focus on AD detection incorporating Support Vector Machines, while deep learning techniques have not yet been applied on large EEG datasets. Promising conclusions for future studies are presented.

The problem of locating the global minimum of a function is a challenging one with application in many problems. A common method to tackle this problem is the so-called multistart method, which is the base method for many modern optimization methods. This article proposes a new sampling technique for multistart-based methods, that utilizes an artificial neural network as an approximator of the original objective function. The proposed sampling technique is tested against uniform sampling on a wide set of well-known benchmark optimization problems from the relevant literature and the results are reported.

The purpose of fetal monitoring during childbirth is the early recognition of any pathological conditions to guide a clinician in early intervention to avoid any complication in the health of the fetus. Non-Invasive Fetal Electrocardiography (NIFECG) represents an alternative fetal monitoring technique. The fetal ECG (fECG) derived from maternal thoracic and abdominal ECG recordings, provides an alternative to typical embryo monitoring means. In addition, it allows for long-term and ambulatory registrations that broaden the diagnostic capabilities for assessing the fetal health. However, in real situations, clear fECG is difficult to extract because it is usually overwhelmed by the dominant maternal ECG and other contaminated noise such as baseline wander and high-frequency interference. In this paper, a novel integrated adaptive methodology based on the combination of blind source separation, empirical mode decomposition, wavelet shrinkage denoising and correlation analysis, for the non-invasive extraction and processing of the FECG, is proposed. The methodology has been evaluated using both real and simulated recordings, and the obtained results indicate it efficiently.

In this manuscript, a methodology for analysing motor signals from Parkinson’s disease (PD) patients is presented. The signals are obtained from PD patients while wearing a glove device and sequentially performing standard motor tests. The signals are processed in order to detect the onset and offset from specific items (items 23-25) of the Unified Parkinson’s Disease Rating Scale (UPDRS) and then the isolated signal parts are analysed in order to quantity the motor findings defined in UPDRS for these items, such as hesitation, movement amplitude and frequency, and rotation range. The obtained results indicate that the methodology can achieve accurate motor assessment (related to ground-truth UPDRS) for both “Off” and “On” stages.

Prolonging the lifetime of a wireless sensor network (WSN) is closely related to the energy hole problem where nodes close to the sink node consume more energy than the other nodes, due to their increased traffic load. For tackling the effects of the energy hole problem, a framework is proposed here consisting of two parts. In its first part, the sink node is selected in order to minimize the energy consumption of the most severely affected by the energy hole problem node. During the second part, a new algorithm proposed here is implemented and changes the path followed by traffic forwarded by nodes close to the sink node in order to achieve less energy consumption for the most energy consuming node of the network.

Nonalcoholic fatty liver disease (NAFLD) is one of the most frequent liver conditions representing a wide range of intrahepatic disorders, varying from steatosis to nonalcoholic steatohepatitis (NASH). Steatosis refers to the accumulation of benign fat cells, which at higher rates leads to NASH progression, as the major risk factor for hepatic fibrosis and cirrhosis, as well as for hepatocellular carcinoma (HCC). In recent years the medical field has focused on preventing the progression of these diseases, with microscopic biopsy images being the gold standard imaging modality in modern clinical trials. The proposed work aims at the high classification ability of four histological liver structures, by training a convolutional neural network (CNN) and comparing its diagnostic performance with various pre-trained deep CNN architectures. All diagnostic attempts were made on an augmented image dataset, with the new CNN model achieving a 95.8% classification accuracy, while AlexNet emerging as the most efficient architecture with a corresponding performance of 97.8%.

This paper presents a hybrid algorithm for the creation of heterogeneous single layer neural networks (SLNNs). The proposed self-adaptive heterogeneous hybrid extreme learning machine (SA-He-HyELM) trains a series of SLNNs with different neuron types in the hidden layer utilizing the extreme learning machine (ELM) algorithm. These networks are evolved into heterogeneous networks (networks having different combinations of hidden neurons) with the help of a modified genetic algorithm (GA). The algorithm is able to handle two architecturally different neuron types: traditional low order (linear) units and higher order units with different transfer functions. The GA is fully self-adaptive and uses one novel hybrid crossover operator along with a self-adaptive mutation operator in order to retain ELM’s simplicity and minimize the number of parameters need tuning. The experimental part of the current paper involves testing SA-He-HyELM with traditional ELM and other three ELM-based methods. The experimental part utilized a series of regression and classification experiments on relatively large datasets. In all cases the proposed method managed to get lower MSE or higher classification accuracy when compared to the aforementioned methods.

Colorectal cancer (CRC) is the second most common cancer in women and the third most common in men, with an increasing incidence. Pathology diagnosis complemented with prognostic and predictive biomarker information is the first step for personalized treatment. The increased diagnostic load in the pathology laboratory, combined with the reported intra- and inter-variability in the assessment of biomarkers, has prompted the quest for reliable machine-based methods to be incorporated into the routine practice. Recently, Artificial Intelligence (AI) has made significant progress in the medical field, showing potential for clinical applications. Herein, we aim to systematically review the current research on AI in CRC image analysis. In histopathology, algorithms based on Deep Learning (DL) have the potential to assist in diagnosis, predict clinically relevant molecular phenotypes and microsatellite instability, identify histological features related to prognosis and correlated to metastasis, and assess the specific components of the tumor microenvironment.

Non-alcoholic fatty pancreas disease (NAFPD) is a common and at the same time not extensively examined pathological condition that is significantly associated with obesity, metabolic syndrome, and insulin resistance. These factors can lead to the development of critical pathogens such as type-2 diabetes mellitus (T2DM), atherosclerosis, acute pancreatitis, and pancreatic cancer. Until recently, the diagnosis of NAFPD was based on noninvasive medical imaging methods and visual evaluations of microscopic histological samples. The present study focuses on the quantification of steatosis prevalence in pancreatic biopsy specimens with varying degrees of NAFPD. All quantification results are extracted using a methodology consisting of digital image processing and transfer learning in pretrained convolutional neural networks for the detection of histological fat structures. The proposed method is applied to 20 digitized histological samples, producing an 0.08% mean fat quantification error thanks to an ensemble CNN voting system and 83.3% mean Dice fat segmentation similarity compared to the semi-quantitative estimates of specialist physicians.

Drug inefficiency in patients with refractory seizures renders epilepsy a life-threatening and challenging brain disorder and stresses the need for accurate seizure detection and prediction methods and more personalized closed-loop treatment systems. In this paper, a multicenter methodology for automated seizure detection based on Discrete Wavelet Transform (DWT) is presented. A decomposition of 5 levels is applied in each EEG segment and five features are extracted from the wavelet coefficients. The extracted feature vector is used to train a Random Forest classifier and discriminate between ictal and interictal data. EEG recordings from the database of University of Bonn and the database of the University Hospital of Freiburg were employed, in an attempt to test the efficiency and robustness of the method. Classification results in both databases are significant, reaching accuracy above 95% and confirming the robustness of the methodology. Sensitivity and False Positive Rate for the Freiburg database reached 99.74% and 0.21/h respectively.

A genetic programming tool is proposed here for data classification. The tool is based on Grammatical Evolution technique and it is designed to exploit multicore computing systems using the OpenMP library. The tool constructs classification programs in a C-like programming language in order to classify the input data, producing simple if-else rules and upon termination the tool produces the classification rules in C and Python files. Also, the user can use his own Backus Normal Form (BNF) grammar through a command line option. The tool is tested on a wide range of classification problems and the produced results are compared against traditional techniques for data classification, yielding very promising results.

Dear Editor-in-Chief, Previous letters have suggested the implementation of certain measures to either ameliorate the effects of the financial crisis or even modernize the Greek academic institutions over the previous decade (2010-2019), either in general1 or in the context of biomedical education.2 In these letters, the authors proposed a series of predesigned steps to achieve the desired results. What we would like to raise the attention of the readership to, is another driving force of change in the Greek academic sector (other than carefully designed plans): forced adaptation to external circumstances, like to COVID-19 pandemic. Some European countries (like the United Kingdom or Nordic countries) have made serious investments in e-learning platforms and curricula over the last 10-15 years (and thus acclimatized their corresponding academic communities into the concept of e-exams),3 while others, like Greece, are not similarly mature in efficiently (from an organizational, pedagogic and technological point of view) integrating distant learning approaches and e-examinations into their academic curricula. Read more...

A modified version of a common global optimization method named controlled random search is presented here. This method is designed to estimate the global minimum of multidimensional symmetric and asymmetric functional problems. The new method modifies the original algorithm by incorporating a new sampling method, a new termination rule and the periodical application of a local search optimization algorithm to the points sampled. The new version is compared against the original using some benchmark functions from the relevant literature.

Non-alcoholic fatty liver disease (NAFLD) covers a range of chronic medical conditions varying from hepatocellular inflammation which characterizes nonalcoholic steatohepatitis (NASH) to steatosis, as the key element of a nonalcoholic fatty liver (NAFL). It is globally linked to the increasing prevalence of obesity and other components of metabolic syndrome and is expected to be the main indication for the existence of the liver disease in the coming years. Its eradication has become a major challenge due to the difficulties in clinical diagnosis, complex pathogenesis and the lack of approved therapies. In this study, an automated image analysis method is presented providing quantitative assessments of fat deposition in steatotic liver biopsy specimens. The methodology applies image processing, machine learning and evolutionary algorithm optimization techniques, producing a 1.93% mean classification error compared to the semiquantitative interpretations of specialized hepatologists.

An increasing body of evidence suggests that antipsychotic medication can cause immunological changes that could be attributed to the amelioration of psychotic symptoms or the metabolic side effects of the drugs. So far, the results of the studies remain controversial. Our aim was to compare the levels of interleukin (IL) IL-2, IL-6 and transforming growth factor-β2 (TGF-β2) in drug-naive, first-episode patients with psychosis before and after six weeks of antipsychotic medication. Thirty-nine first-episode patients with psychosis were enrolled in the study. Serum levels of IL-2, IL-6 and TGF-β2 were measured by enzyme linked immunosorbent assay (ELISA) before and six weeks after the initiation of antipsychotics. In addition, clinical psychopathology was assessed using Positive and Negative Syndrome Scale (PANSS) before and after treatment. Serum levels of IL-2 were significantly increased six weeks after the initiation of antipsychotic treatment ( p 0.001 ) while TGF-β2 levels were decreased ( p 0.001 ). IL-6 levels were overall increased ( p 0.004 ), but this occurred in a non-linear way. These findings, although preliminary, provide further evidence that antipsychotic treatment in patients with psychosis may be correlated with immunological changes but further research is needed.

Background & Aims Liver biopsy is the reference standard for staging and grading nonalcoholic fatty liver disease (NAFLD), but histologic scoring systems are semiquantitative with marked interobserver and intraobserver variation. We used machine learning to develop fully automated software for quantification of steatosis, inflammation, ballooning, and fibrosis in biopsy specimens from patients with NAFLD and validated the technology in a separate group of patients. Methods We collected data from 246 consecutive patients with biopsy-proven NAFLD and followed up in London from January 2010 through December 2016. Biopsy specimens from the first 100 patients were used to derive the algorithm and biopsy specimens from the following 146 were used to validate it. Biopsy specimens were scored independently by pathologists using the Nonalcoholic Steatohepatitis Clinical Research Network criteria and digitalized. Areas of steatosis, inflammation, ballooning, and fibrosis were annotated on biopsy specimens by 2 hepatobiliary histopathologists to facilitate machine learning. Images of biopsies from the derivation and validation sets then were analyzed by the algorithm to compute percentages of fat, inflammation, ballooning, and fibrosis, as well as the collagen proportionate area, and compared with findings from pathologists’ manual annotations and conventional scoring systems. Results In the derivation group, results from manual annotation and the software had an interclass correlation coefficient (ICC) of 0.97 for steatosis (95% CI, 0.95–0.99; P < .001); ICC of 0.96 for inflammation (95% CI, 0.9–0.98; P < .001); ICC of 0.94 for ballooning (95% CI, 0.87–0.98; P