Archive for the ‘Machine Learning’ Category

Here’s How Companies are Using AI, Machine Learning – Dice Insights

Companies widely expect that artificial intelligence (A.I.) and machine learning will fundamentally change their operations in coming years. To hear executives talk about it, apps will grow smarter, tech stacks will automatically adapt to vulnerabilities, and processes throughout organizations will become entirely automated.

Given the buzz around A.I., its easy for predictions to easily slip into the realm of the fantastical (In less than six months, well have cars that drive themselves! And apps that predict what a user wants before they want it!). Its worth taking a moment to see what companies areactuallydoing with A.I. at this juncture.

To that end, CompTIArecently asked 400 companiesabout their most common use-cases for A.I. Heres what they said:

The pandemic has accelerated digital transformation and changed how we work, Khali Henderson, Senior Partner at BuzzTheory and vice chair of CompTIAs Emerging Technology Community, wrote in a statement accompanying the data.We learnedsomewhat painfullythat traditional tech infrastructure doesnt provide the agility, scalability and resilience we now require. Going forward, organizations will invest in technologies and services that power digital work, automation and human-machine collaboration. Emerging technologies like AI and IoT will be a big part of that investment, which IDC pegs at $656 billion globally this year.

That predictive sales/lead scoring would top this list makes a lot of senseif companies are going to invest in A.I., theyre likely to start with a process that can provide a rapid return on investment (and generate a lot of cash).According to CompTIA, A.I. helps with more effective prioritization of sales prospects via lead scoring and provides detailed, real-time analytics. Its a similar story with CRM/service delivery optimization, where A.I. can help salespeople and technologists better identify potential customers and cross-selling opportunities.

Companies have spent years working on chatbots and digital assistants, hoping that automated systems can replace massive, human-powered call centers. So far, theyve had mixed results;the early generations of chatbotswere capable of conducting simple interactions with customers, but had a hard time with complex requests and the nuances of language. The emergence of more sophisticated systems likeGoogle Duplexpromises a future in which machines effectively chat with customers on a range of issuesprovided customers can trust interacting with software in place of a human being.

As A.I. and machine learning gradually evolve, opportunities to work with the technology will increase. While many technologists tend to equate artificial intelligence withcutting-edge projectssuch as self-driving cars, this CompTIA data makes it clear that companies first use of A.I. and machine learning will probably involve sales and customer service. Be prepared.

Membership has its benefits. Sign up for a free Dice profile, add your resume, discover great career insights and set your tech career in motion. Register now

See original here:
Here's How Companies are Using AI, Machine Learning - Dice Insights

Column: Simplifying live broadcast operations using AI and machine learning – NewscastStudio

Subscribe toNewscastStudio's newsletter for the latest in broadcast design and engineering delivered to your inbox.

Artificial Intelligence and machine learning are seen as pillars of the next generation of technological advancement in broadcast media for a variety of reasons, including the ability to sift through mountains of data while identifying anomalies, spotting trends and alerting users to potential problems before they occur without the need for human intervention. With the more data they ingest these models improve over time, meaning the more ML models utilized across a variety of applications, the faster and more complex the insights derived from these tools become.

But to truly understand why machine learning provides enormous value for broadcasters, lets break it down into use cases and components within broadcast media where AI and ML can have the greatest impact.

Imagine a live sporting event stopsstreaming,or that framesstart dropping for no apparent reason.Viewers are noticing quality problems and starting to complain.Technicians are baffled and customers may have just missed the play of the year. Revenue therefore takes a hit and executives want to know what is to blame.

These are situations every broadcaster wants to avoid, and in these tense moments there is no time to lose viewers are flipping to otherservices andad revenue is being lost by the second. What went wrong? Who or what is to blame and how can we get this back up and running immediately, while mitigating this risk in the future? Modern broadcasters need to know before problems happen not be caught in a crisis trying to pick up the pieces after an incident.

Advertisement

The promise of our interconnected world means video workflowsareinteracting, intertwining, and integrating in new ways every day, simultaneously increasing information sharing, agility and connectivity while producing increasingly complex challenges and issues to diagnose. As more on-prem and cloud resources are connected with equipment from different vendors, sources, and partner organizationsdistributing to new device types,thereisan enormous, ever-expanding number of log and telemetrydata produced.

As a result, broadcastengineers have more information than they can effectively process. They routinely silence frequent alerts and alarms because with too much data overload it can be impossible to tellwhat isimportant and what is not. This inevitably leaves teams overwhelmed and lacking insights.

Advanced analytics and ML can help with these problems by making sense of overwhelming quantities of data, allowing human operators to sift through insignificant clutter and to focus and understand where issues are likely to occur before failures are noticed. Advanced analytics provide media companies the unprecedented opportunity to leverage sophisticated event correlation, data aggregation, deep learning, and virtually limitless applications to improve broadcast workflows. The benefit is to be able to do more with less, to innovate faster than the competition and prepare for the future both by increasing your knowledge base and opening the potential for cost reduction and time savings, honing in on the crucial details behind the data that matters most to both their users and organization.

One of the biggest challenges facing broadcast operations engineers is to recognize when things are not working before the viewers experience is affected. In a perfect world operators and engineers want to predict outages and identify potential issues ahead of time. Machine learning models can be orchestrated to recognize the normal ranges based on hundreds to thousands of measurements beyond the ability of a human operator and alert the operator in real time when a stream anomaly occurs. While this process normally requires monitoring logs on dozens of machines and keeping track of the performance of network links between multiple locations and partners, using ML allows the system to identify patterns in large data sets and helps operators focus only on workflow anomalies dramatically reducing workload.

Anomaly detection works by building a predictive model of what the next measurements related to a stream will be for example, the round-trip time of packets on the network or the raw bitrate of the stream and then determining how different the expected value is from the next measurement. As a tool to sort through normal and abnormal streams, this can be essential, especially when managing hundreds or thousands of concurrent channels. One benefit of anomalous behavior identification would be enabling an operator to switch to a backup link that uses a different network link before a failure occurs.

Anomaly detection can also be a vital component of reducing needless false alarms and reducing time waste. Functionality such as customizable alerting preferences and aggregated health scores generated by threat-gauging data points assist operators to sift through and assimilate data trends so they can focus where they really need to. In addition, predictive and proactive alerting can be orders of magnitude less expensive and allow broadcasters to be able to identify the root causes of instability and failure faster and easier.

A major challenge to any analytics system is data collection. When you have a video workflow comprised of machines in disparate data centers running different operating systems and tools, it can be difficult to assimilate and standardize reliable, relevant data that can be used in any AI/ML system. While there are natural data aggregation points in most broadcast architectures for example if you are using a cloud operations and remote management platform or common protocol stack this is certainly not a given.

Although standards exist for how video data should be formatted and transmitted, few actually describe how machine data, network measurements, and other telemetry should be collected, transmitted and stored. Therefore it is essential to select a technology partner that sends data to a common aggregation point where it is parsed, normalized and put into a database while supporting multiple protocols to support a robust AI/ML solution.

Once you have a method for collecting real-time measurements from your video workflow, you can feed this data into a ML engine to detect patterns. From there you can train the system not only to understand normal operating behavior for anomaly detection, but also to recognize specific patterns leading up to video degradation events. With these patterns determined you can also identify common metadata related to degradation events across systems, allowing you to identify that the degradation event is related to a particular shared network segment.

For example, if a particular ISP in a particular region continues to experience latency or blackout issues, the system learns to pick up on warning signs ahead of time and notifies the engineer before an outage preventing issues proactively while simultaneously improving root cause identification within your entire ecosystem. Developers can also see that errors are more often observed using common encoder or network hardware settings. Unexpected changes in the structure of the video stream or the encoding quality might also be important signals of impending problems. By observing correlations, ML gives operators key insights into the causes of problems and how to solve them.

Predictive analytics, alerts and correlations are useful for automated failure prediction and alerting, but when all else fails, ML models can also be used to help operators concentrate on areas of concern following an outage, making retrospective analysis much easier and faster via root cause analysis.

With workflows that consist of dozens of machines and network segments, it is inherently difficult to know where to look for problems. However, ML models, as we have seen, provide trend identification and help visualize issues using data aggregation. Even relatively straightforward visualizations of how a stream deviates from the norm are incredibly valuable, whether in the form of historical charts, customizable reports or questions as simple as how a particular stream compares to a similar recent stream.

Leveraging AI and ML to improve operational efficiency and quality provides a powerful advantage while preparing broadcasters for the future of live content delivery over IP. Selecting the right vendor for system monitoring and orchestration that integrates AI and ML capabilities can help your organization make sense of the vast amounts of data being sent across the media supply chain and be a powerful differentiator.

As experiments to test hypotheses are essential to the traditional learning process, the same goes for ML models. Building, training, deploying, and updating ML models are inherently complex, meaning providers in cooperation with their users must continue to iterate, compare results, and adjust accordingly to understand the why behind the data, improving root cause analysis and the customer experience.

Machine learning presents an unprecedented opportunity for sophisticated event correlation, data aggregation, deep learning, and virtually unlimited applications across broadcast media operations as it evolves exponentially year to year. As models become more informed and interconnected, problem solving and resolution technology based on deep learning and AI will become increasingly essential tools. Broadcast organizations looking to prepare themselves for such a future would be wise to prepare for this eventuality by choosing the right vendor to integrate AI and ML enabled tools into their workflows.

Andrew leads Zixis Intelligent Data Platform initiative, bringing AI and ML to live broadcast operations. Before Zixi he led the video platform product team at Brightcove where he spent 6 years working with some of the largest broadcasters and media companies. Particular areas of interest include live streaming, analytics, ad integration, and video players. Andrew has an MBA from Babson College and a BA from Oberlin College.

Continue reading here:
Column: Simplifying live broadcast operations using AI and machine learning - NewscastStudio

Deep machine learning study finds that body shape is associated with income – PsyPost

A new study published in PLOS One has found a relationship between a persons body shape and their family income. The findings provide more evidence for the beauty premium a phenomenon in which people who are physically attractive tend to earn more than their less attractive counterparts.

Researchers have consistently found evidence for the beauty premium. But Suyong Song, an associate professor at The University of Iowa, and his colleagues observed that the measurements used to gauge physical appearance suffered some important limitations.

I have been curious of whether or not there is physical attractiveness premium in labor market outcomes. One of the challenges is how researchers overcome reporting errors in body measures such as height or weight, as most previous studies often defined physical appearance from subjective opinions based on surveys, Song explained.

The other challenge is how to define body shapes from these body measures, as these measures are too simple to provide a complete description of body shapes. In this study, collaborated with one of my coauthors (Stephen Baek at University of Virginia), we use novel data which contains three-dimensional whole-body scans. Using a state-of-the art machine learning technique, called graphical autoencoder, we addressed these concerns.

The researchers used the deep machine learning methods to identify important physical features in whole-body scans of 2,383 individuals from North America.

The data came from the Civilian American and European Surface Anthropometry Resource (CAESAR) project, a study conducted primarily by the U.S. Air Force from 1998 to 2000. The dataset included detailed demographic information, tape measure and caliper body measurements, and digital three-dimensional whole-body scans of participants.

The findings showed that there is a statistically significant relationship between physical appearance and family income and that these associations differ across genders, Song told PsyPost. In particular, the males stature has a positive impact on family income, whereas the females obesity has a negative impact on family income.

The researchers estimated that one centimeter increase in stature (converted in height) is associated with approximately $998 increase in family income for a male who earns $70,000 of the median family income. For women, the researchers estimated that one unit decrease in obesity (converted in BMI) is associated with approximately $934 increase in the family income for a female who earns $70,000 of family income.

The results show that the physical attractiveness premium continues to exist, and the relationship between body shapes and family income is heterogeneous across genders, Song said.

Our findings also highlight importance of correctly measuring body shapes to provide adequate public policies for improving healthcare and mitigating discrimination and bias in the labor market. We suggest that (1) efforts to promote the awareness of such discrimination must occur through workplace ethics/non-discrimination training; and (2) mechanisms to minimize the invasion of bias throughout hiring and promotion processes, such as blind interviews, should be encouraged.

The new study avoids a major limitation of previous research that relied on self-reported attractiveness and body-mass index calculations, which do not distinguish between fat, muscle, or bone mass. But the new study has an important limitation of its own.

One major caveat is that the data set only includes family income as opposed to individual income. This opens up additional channels through which physical appearance could affect family income, Song explained. In this study, we identified the combined association between body shapes and family income through the labor market and marriage market. Thus, further investigations with a new survey on individual income would be an interesting direction for the future research.

The study, Body shape matters: Evidence from machine learning on body shape-income relationship, was published July 30, 2021.

View original post here:
Deep machine learning study finds that body shape is associated with income - PsyPost

Apple’s Machine Learning Research Team have Published a Paper on using Specialized Health Sensors in Future AirPods – Patently Apple

Apple began discussing integrating health sensors into future sports-oriented headphones in a patent application that was published back in April 2009 and filed in 2008. Apple's engineers noted at the time that "The sensor can also be other than (or in addition to) an activity sensor, such as a psychological or biometric sensors which could measure temperature, heartbeat, etc. of a user of the monitoring system." Fast forwarding to 2018, Apple decided to update their AirPods trademark by adding "wellness sensors" to its description, a telltale sign something was in-the-works. Then a series of patents surfaced in 2020-21 timeline covering health sensor for future AirPods (01,02&03). To top it all off, in June of this year, Apple's VP of Technology talked about health sensors on Apple Watch and possibly AirPods.

The latest development on this front came from Apple's Machine Learning (ML) Research team earlier this month in the form of a research paper. Apple notes, "In this paper, we take the first step towards developing a breathlessness measurement tool by estimating respiratory rate (RR) on exertion in a healthy population using audio from wearable headphones. Given this focus, such a capability also offers a cost-effective method to track cardiorespiratory fitness over time. While sensors such as thermistors, respiratory gauge transducers, and acoustic sensors provide the most accurate estimation of a persons breathing patterns, they are intrusive and may not be comfortable for everyday use. In contrast, wearable headphones are relatively economical, accessible, comfortable, and aesthetically acceptable."

Further into the paper, Apple clarifies: "All data was recorded using microphone-enabled, near-range headphones, specifically Apples AirPods. These particular wearables were selected because they are owned by millions and utilized in a wide array of contexts, from speaking on the phone to listening to music during exercise."

(Click on image to greatly Enlarge)

Below is a full copy of the research paper published by Apple's Machine Learning Research team in the form of a SCRBD document, courtesy of Patently Apple.

Machine Learning Team Paper on Respiratory Rates in Wearable Microphones by Jack Purcher on Scribd

While the paper doesn't discuss when these specialized sensors using machine learning techniques will be implemented in AirPods, it's clearly a positive development that Apple is well into the process of proving the value of adding such sensors to future AirPods.

Read the original post:
Apple's Machine Learning Research Team have Published a Paper on using Specialized Health Sensors in Future AirPods - Patently Apple

Machine learning to improve prognosis prediction of EHCC | JHC – Dove Medical Press

Introduction

Hepatocellular carcinoma (HCC), the fourth leading cause of cancer-related death worldwide, typically occurs in patients with chronic liver disease and is an aggressive disease with dismal prognosis.1 Over the past decades, improved surveillance programs and imaging techniques have led to early HCC (EHCC) diagnosis in 4050% of patients, at a stage amenable to potentially curative therapiesresection, transplantation or ablation.2,3 Generally, EHCC is expected to have an excellent outcome after radical therapies. Since total hepatectomy eliminates both the diseased liver and the tumor, liver transplantation (LT) offers the highest chance of cure, with a survival up to 70% at 10 years in selected cases, and remains the best treatment for EHCC.4 Unfortunately, the critical shortage of donor organs represents the main limitation of LT and results in long waiting times.

According to clinical practice guidelines, liver resection (LR) is the recommended first-line option for patients with EHCC and preserved liver function, although ablation is an alternative treatment modality.3,5,6 The prognosis following LR may vary even among patients with EHCC and two competing causes of death (tumor recurrence and liver dysfunction) both influence survival.7 Several HCC staging systems have been proposed to pair prognostic prediction with treatment allocation; however, these proposalssuch as Barcelona Clinic Liver Cancer (BCLC) staging, China Liver Cancer (CNLC) staging, Hong Kong Liver Cancer (HKLC) staging and Cancer of the Liver Italian Program (CLIP) scoreare not derived from surgically managed patients, except for the American Joint Committee on Cancer (AJCC) system and Japan Integrated Staging (JIS) score, and therefore exhibit modest prognostic accuracy for resected cases.69 A few prognostic models have been developed based on readily available patient and tumor characteristics; however, they are by nature outmoded and rigid tools because all determinants were examined by conventional statistical methods (ie, Cox proportional hazard regression) and assigned fixed weights.8,10 Hence, new strategies to improve outcome prediction and treatment selection are warranted for EHCC patients.

Machine learning (ML), a subfield of artificial intelligence, leverages algorithmic methods that enable computers to learn from on large-scale, heterogeneous datasets and execute a specific task without predefined rules.11 ML solutions such as gradient boosting machine (GBM) have outperformed regression modelling in a variety of clinical situations (eg, diagnosis and prognosis).1113 Nevertheless, the benefit of ML in predicting prognosis of patients with resected EHCC has yet to be fully explored. Accordingly, we assembled a large, international cohort of EHCC patients to design and evaluate a ML-based model for survival prediction, and compare its performance with existing prognostic systems.

Patients with EHCC, defined as tumor 5 cm and without evidence of extrahepatic disease or major vascular invasion,14 were retrospectively screened from two sources: (1) Medicare patients treated with surgical therapy (LR+LT) in the Surveillance, Epidemiology, and End Results (SEER) Program, a population-based database in the United States, between 2004 and 2015; (2) consecutive patients treated with LR at two high-volume hepatobiliary centers in China (First Affiliated Hospital of Nanjing Medical University and Wuxi Peoples Hospital) between 2006 and 2016. The inclusion criteria were (1) adult patients aged 20 years; (2) histology-confirmed HCC (International Classification of Diseases for Oncology, Third Edition, histology codes 8170 to 8175 for HCC and site code C22.0 for liver);15 (3) complete survival data and a survival of 1 month. The exclusion criteria were (a) missing information on the type of surgical procedure; (b) another malignant primary tumor prior to HCC diagnosis; (c) unknown cause of death. Patient selection process is summarized in the flow chart of Figure 1. This study protocol was approved by the Institution Review Board of First Affiliated Hospital of Nanjing Medical University and Wuxi Peoples Hospital. Written informed consent was waived because retrospective anonymous data were analyzed. Non-identified information was used in order to protect patient data confidentiality. This study was conducted in accordance with the Declaration of Helsinki.

Figure 1 Analytical framework for survival prediction. (A) Flow diagram of the study cohort details. (B) A machine learning pipeline to train, validate and test the model.

The endpoint selected to develop ML-based model was disease-specific survival (DSS), defined as the time from the date of surgery to the date of death from disease (tumor relapse or liver dysfunction). All deaths from any other cause were counted as non-disease-specific and censored at the date of the last follow-up. Follow-up protocol for Chinese cohort included physical examination, laboratory evaluation and dynamic CT or MRI of the chest and abdomen every 3 months during the first 2 years and every 6 months thereafter. The follow-up was terminated on August 15, 2020.

Electronic and paper medical records were reviewed in detail; all pertinent demographic and clinicopathologic data were abstracted on a standardized template. The following characteristics of interest were ascertained at the time of enrollment: age, gender, race, year of diagnosis, alpha-fetoprotein level, use of neoadjuvant therapy, tumor size, tumor number, vascular invasion, histological grade, liver fibrosis score, and type of surgery.

We deployed GBM, a decision tree-based ML algorithm that has gained popularity because of its performance and interpretability, to aggregate baseline risk factors and predict the likelihood of survival using the R package gbm. GBM algorithm16 assembles multiple base learners, in a step-wise fashion, with each successive learner fitting the residuals left over from previous learners to improve model performance: (1) , where is a base learner, typically a decision tree; (2) , where is optimized parameters in each base learner and is the weight of each base learner in the model. Each base learner may have different variables; variables with higher relative importance are utilized in more decision trees and earlier in the boosting algorithm. The model was trained using stratified 33-fold nested cross-validation (3 outer iterations and 3 inner iterations) on the training/validation cohort; a grid search of optimal hyper-parameter settings was run using the R package mlr. Figure 1 shows the ML workflow schematically.

Model discrimination was quantified using Harrells C-statistic and 95% confidence intervals [CIs] were assessed by bootstrapping. Calibration plots were used to assess the model fit. Decision curve analysis was used to determine the clinical net benefit associated with the adoption of the model.17

Differences between groups were tested using 2 test for categorical variables and MannWhitney U-test for continuous variables. Survival probabilities were assessed using the KaplanMeier method and compared by the Log rank test. The optimal cutoffs of GBM predictions were determined to stratify patients at low, intermediate, or high risk for disease-specific death by using X-tile software version 3.6.1 (Yale University School of Medicine, New Haven, CT).18 Propensity score matching (PSM) was used to balance the LR versus LT for EHCC in SEER cohort using 1:1 nearest neighbor matching with a fixed caliper width of 0.02. Cases (LR) and controls (LT) were matched on all baseline characteristics other than type of surgery using the R package MatchIt. All analyses were conducted using R software version 3.4.4 (www.r-project.org). Statistical significance was set at P<0.05; all tests were two-sided.

A total of 2778 EHCC patients (2082 males and 696 females; median age, 60 years; interquartile range [IQR], 5467 years) treated with LR were identified and divided into 1899 for the training/validation (SEER) cohort and 879 for the test (Chinese) cohort. Patient characteristics of the training/validation and test cohorts are summarized in Table 1. There were 625 disease-related deaths recorded (censored, 67.1%) during a median (IQR) follow-up time of 44.0 (26.074.0) months in the SEER cohort, and 258 deaths were recorded (censored, 70.6%) during a median (IQR) follow-up of 52.5 (35.876.0) months in the Chinese cohort. Baseline characteristics and post-resection survival differed between the cohorts.

Table 1 Baseline Characteristics in the Training/Validation and Test Cohorts

We investigated 12 potential model covariates using GBM algorithm. According to the results of nested cross-validation, we utilized 2000 decision trees sequentially, with at least 5 observations in the terminal nodes of the trees; the decision tree depth was optimized at 3, corresponding to 3-way interactions, and the learning rate was optimized at 0.01. Covariates with a relative influence greater than 5 (age, race, alpha-fetoprotein level, tumor size, multifocality, vascular invasion, histological grade and fibrosis score) were integrated into the final model developed to predict DSS (Figure 2A and B).

Figure 2 Overview of the machine-learning-based model. (A) Relative importance of the variables included in the model. (B) Illustrative example of the gradient boosting machine (GBM). GBM builds the model by combining predictions from stumps of massive decision-tree-base-learners in a step-wise fashion. GBM output is calculated by adding up the predictions attached to the terminal nodes of all 2000 decision trees where the patient traverses. (C) Performance of GBM model as compared with that of American Joint Committee on Cancer (AJCC) staging in the internal validation group. (D) Online model deployment based on GBM output.

The final GBM model demonstrated good discriminatory ability in predicting post-resection survival specific for EHCC, with a C-statistic of 0.738 (95% CI 0.7170.758), and outperformed the 7th and 8th edition of AJCC staging systems (P<0.001) in the training/validation cohort (Table 2). The internal validation group was the 33-fold nested cross-validation of the final model of the training cohort with 211 patients in each fold. For the composite outcome, the GBM model yielded a median C-statistic of 0.727 (95% CI 0.7060.761) and performed better than AJCC staging systems (P<0.05) in the internal validation group (Figure 2C). In the test cohort, the GBM model provided a C-statistic of 0.721 (95% CI, 0.6890.752) in predicting DSS after resection of EHCC and was clearly superior to AJCC, BCLC, CNLC, HKLC, CLIP and JIS systems (P<0.05). Note that prediction scores differed between training/validation and test sets (P<0.001) (Figure S1). The discriminatory performance of ML-based model exceeded those of AJCC staging systems even in sub-cohorts stratified by covariate integrity (complete/missing) (Table S1). Furthermore, the GBM model exhibited greater ability to discriminate survival probabilities than simple prognostic strategies, such as multifocal EHCC with vascular invasion indicating a dismal prognosis following LR, in sub-cohorts with complete strategy-related information (P<0.001) (Table S2).

Table 2 Performance of GBM Model and Staging Systems

Calibration plots presented excellent agreement between model predicted and actual observed survival in both the training/validation and test cohorts (Figure S2A and B). Decision curve analysis demonstrated that the GBM model provided better clinical utility for EHCC in designing clinical trials than the treat all or treat none strategy across the majority of the range of reasonable threshold probabilities (Figure S2C and D). The model is publicly accessible for use on Github (https://github.com/radgrady/EHCC_GBM), with an app (https://mlehcc.shinyapps.io/EHCC_App/) that allows survival estimates at individual scale (Figure 2D).

We utilized X-tile analysis to generate two optimal cut-off values (6.35 and 5.32 in GBM predictions, Figure S3) that separated EHCC patients into 3 strata with a highly different probability of post-resection survival in the training/validation cohort: low risk (760 [40.0%]; 10-year DSS, 75.6%), intermediate risk (948 [49.9%]; 10-year DSS, 41.8%), and high risk (191 [10.1%]; 10-year DSS, 5.7%) (P<0.001). In the test cohort, the aforementioned 3 prognostic strata by using the GBM model were confirmed: low risk (634 [72.1%]; 10-year DSS, 69.0%), intermediate risk (194 [22.1%]; 10-year DSS, 37.9%), and high risk (51 [5.8%]; 10-year DSS, 4.7%) (P<0.001) (Table 3). Visual inspection of the survival curves again revealed that, compared with the 8th edition AJCC criteria, the GBM model provided better prognostic stratification in both the training/validation and test cohorts (Figure 3). Differences in the baseline patient characteristics according to risk groups defined by the GBM model are summarized in Table S3.

Table 3 Disease-Specific Survival According to Risk Stratification

Figure 3 Kaplan-Meier survival plots demonstrating disparities between groups. Disease-specific survival stratified by the 8th edition of the American Joint Committee on Cancer T stage and the machine-learning model in the training/validation (A and C) and the test (B and D) cohort.

We also gathered data of 2124 EHCC patients (1671 males and 453 females; median age, 58 years; IQR, 5362 years) treated with LT from the SEER-Medicare database. SEER data demonstrated that considerable differences existed between LR (n=1899) and LT (n=2124) cohorts in terms of all listed clinical variables except for alpha-fetoprotein level (Table S4). Upon initial analysis, we found a remarkable survival benefit of LT over LR for patients with EHCC (hazard ratio [HR] 0.342, 95% CI 0.3000.389, P<0.001), which was further confirmed in a well-matched cohort of 1892 patients produced by PSM (HR 0.342, 95% CI 0.2850.410, P<0.001). Although a trend for higher survival probability was observed after 5 years in the LT cohort, no statistically significant difference in DSS was observed when compared with low-risk LR cohort (HR 0.850, 95% CI 0.6791.064, P=0.138). After PSM, 420 patients in the LT cohort were matched to 420 patients in the low-risk LR cohort; the trend for improved survival remained after 5 years in the matched LT cohort while the matched comparison also yielded no significant survival difference (HR 0.802, 95% CI 0.5611.145, P=0.226) (Figure 4). By contrast, when compared with intermediate-and high-risk patients treated with LR, remarkable survival benefits were observed in patients treated with LT both before and after PSM (P<0.001) (Table S5).

Figure 4 Comparison of survival after resection versus transplantation before and after propensity score matching in SEER-Medicare database. (A) KaplanMeier curves for different risk groups stratified by the model in the SEER resection cohort (n=1899) and patients in the SEER transplantation cohort (n=2124). (B) KaplanMeier curves for low-risk patients treated with resection and patients treated with transplantation in propensity score-matched cohort (n=840).

In this study involving over 2700 EHCC patients treated with resection, a gradient-boosting ML model was trained, validated and tested to predict post-resection survival. Our results demonstrated that this ML model utilized readily available clinical information, such as age, race, alpha-fetoprotein level, tumor size and number, vascular invasion, histological grade and fibrosis score, and provided real-time, accurate prognosis prediction (C-statistic >0.72) that outperform traditional staging systems. Among the model covariates, tumor-related characteristics, such as size, multifocality and vascular invasion, as well as liver cirrhosis are known risk factors for poor survival following resection of HCC.710 Besides, multiple population-based studies have shown the racial and age differences in survival of HCC.19,20 Therefore, our ML model is a valid and reliable tool to estimate prognosis of EHCC patients. This study represents, to our knowledge, the first application of a state-of-the-art ML survival prediction algorithm in EHCC based on large-scale, heterogeneous datasets.

In SEER cohort, the 10-year survival rate of EHCC after LR was around 50%, which seemed acceptable but was remarkably lower than that after LT (around 80%). No adjuvant therapies are able to prevent tumor relapse and cirrhosis progression; however, patients with dismal prognosis should be considered candidates for clinical trials of adjuvant therapy.7 Salvage LT has also been a highly applicable strategy to alleviate both graft shortage and waitlist dropout with excellent outcomes that are comparable to upfront LT.1,5 Priority policy, defined as enlistment of patients at high mortality risk before disease progression, was then implemented to improve the transplantability rate.21 Promisingly, our ML tool may help clinicians better identify EHCC patients who are at high risk of disease-related death, engage in clinical trials, and meet priority enlistment policy. Specifically, the GBM model identified 10% of EHCC patients who suffered from extremely dismal prognosis following LR in this study. Given its small proportion and survival benefit, we advocate the pre-emptive enlistment of high-risk subset for salvage LT after LR to avoid the later emergence of advanced disease (ie, tumor recurrence and liver decompensation) ultimately leading to death. Moreover, 40% of EHCC patients were at intermediate risk of disease-related death; adjuvant treatments that target HCC and cirrhosis are desirable. In turn, nearly half of EHCC patients were categorized as low risk by using the GBM model. The low-risk subset permits satisfactory long-term survival after LR and may receive no adjuvant therapy. We note that DSS curves are separated after 5 years for low-risk patients treated with LR as compared with patients treated with upfront LT, and thus long-lasting surveillance should be maintained.

Prior efforts to improve prognostic prediction of EHCC have mostly been reliant on tissue-based or imaging-assisted quantification of research biomarkers.9,22 However, a more accurate, yet more complex, prognosis estimate does not necessarily present a better clinical tool. Parametric regression models are ubiquitous in clinical research because of their simplicity and interpretability; however, regression analysis should be performed in complete cases only.23 Moreover, regression modeling strategies assume that relationships among input variables are linear and homogeneous but complicated interactions exist between predictors.24,25 Decision tree-based methods represent a large family of ML algorithms and can reveal complex non-linear relationships between covariates. GBM algorithm has been widely applied in big data analysis and consistently utilized by the top performers of ML predictive modelling competitions.14,26 GBM algorithm utilizes the boosting procedure to combine stumps of massive decision-tree-base-learners, which is similar to the clinical decision-making process for a patient by aggregating consultations from multiple specialists, each which would that look at the case in a slightly different way. Thus, our GBM model directly integrates interpretability in order to mitigate this issue. Compared with other tree-based ensemble methods such as random forest, GBM algorithm also has a built-in functionality to handle missing values that permits utilizing data from, and assigning classification to, all observations in the cohort without the need to impute data. We applied nested cross-validation scheme for hyperparameter tuning in GBM as it prevents information leaking between observations used for training and validating the model, and estimates the external test error of the given algorithm on unseen datasets more accurately by averaging its performance metrics across folds.27 Comparable discriminatory ability in the training/validation cohort, the test cohort as well as sub-cohorts from different clinical scenarios suggested good reproducibility and reliability of the proposed GBM model.

Our study has several limitations that warrant attention. First, all the presented analyses are retrospective; prospective validations of the ML model in different populations are warranted prior to routine use in clinical practice. Second, the study cohort included population-based cancer registries with limited information regarding patient and tumor characteristics; unavailable confounders, such as biochemical parameters, surgical margin status and recurrence treatment modality could not be adjusted for modeling. Third, SEER-Medicare database contains a considerable amount of missing data in several important clinical variables, such as fibrosis score. Indeed, missing data represent an unavoidable feature of all clinical and population-based databases; however, improper management of data resource, such as simply excluding cases with missing data, can introduce considerable bias, as previously noted across numerous cancer types.28 We therefore contend that integrating missingness into our GBM model indicates good transferability in future clinical practice.

In conclusion, ML approach is both feasible and accurate, and a novel way to consider analysis of survival outcomes in clinical scenarios. Our results suggest that a GBM model trained on readily-available clinical data provides good performance that is better than staging systems in predicting prognosis. Although several issues must be addressed, such as prospective validations and ethical challenges, prior to its widespread use, such an automated tool may complement existing prognostic sources and lead to better personalized treatments for patients with resected EHCC.

EHCC, early hepatocellular carcinoma; LT, liver transplantation; LR, liver resection; BCLC, Barcelona Clinic Liver Cancer; China Liver Cancer, CNLC; HKLC, Hong Kong Liver Cancer; CLIP, Cancer of the Liver Italian Program; AJCC, American Joint Committee on Cancer; ML, machine learning; GBM, gradient boosting machine; SEER, Surveillance, Epidemiology, and End Results; DSS, disease-specific survival; PSM, propensity score matching; IQR, interquartile range.

Data for model training and validation as well as R codes are available at Github (https://github.com/radgrady/EHCC_GBM). Test data are available from the corresponding author (Xue-Hao Wang) on reasonable request.

This study protocol was approved by the Institution Review Board of First Affiliated Hospital of Nanjing Medical University and Wuxi Peoples Hospital. Written informed consent was waived because retrospective anonymous data were analyzed. Non-identified information was used in order to protect patient data confidentiality.

This study was supported by the Key Program of the National Natural Science Foundation of China (31930020) and the National Natural Science Foundation of China (81530048, 81470901, 81670570).

The authors declare no potential conflicts of interest.

1. Yang JD, Hainaut P, Gores GJ, Amadou A, Plymoth A, Roberts LR. A global view of hepatocellular carcinoma: trends, risk, prevention and management. Nat Rev Gastroenterol Hepatol. 2019;16(10):589604. doi:10.1038/s41575-019-0186-y

2. Llovet JM, Montal R, Sia D, Finn RS. Molecular therapies and precision medicine for hepatocellular carcinoma. Nat Rev Clin Oncol. 2018;15(10):599616. doi:10.1038/s41571-018-0073-4

3. European Association for the Study of the Liver. EASL clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol. 2018;69(1):182236. doi:10.1016/j.jhep.2018.03.019

4. Pinna AD, Yang T, Mazzaferro V, et al. Liver transplantation and hepatic resection can achieve cure for hepatocellular carcinoma. Ann Surg. 2018;268(5):868875. doi:10.1097/SLA.0000000000002889

5. Marrero JA, Kulik LM, Sirlin CB, et al. Diagnosis, staging, and management of hepatocellular carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology. 2018;68(2):723750. doi:10.1002/hep.29913

6. Zhou J, Sun H, Wang Z, et al. Guidelines for the diagnosis and treatment of hepatocellular carcinoma (2019 Edition). Liver Cancer. 2020;9(6):682720. doi:10.1159/000509424

7. Villanueva A. Hepatocellular carcinoma. N Engl J Med. 2019;380(15):14501462. doi:10.1056/NEJMra1713263

8. Chan AWH, Zhong J, Berhane S, et al. Development of pre and post-operative models to predict early recurrence of hepatocellular carcinoma after surgical resection. J Hepatol. 2018;69(6):12841293. doi:10.1016/j.jhep.2018.08.027

9. Ji GW, Zhu FP, Xu Q, et al. Radiomic features at contrast-enhanced CT predict recurrence in early stage hepatocellular carcinoma: a Multi-Institutional Study. Radiology. 2020;294(3):568579. doi:10.1148/radiol.2020191470

10. Shim JH, Jun MJ, Han S, et al. Prognostic nomograms for prediction of recurrence and survival after curative liver resection for hepatocellular carcinoma. Ann Surg. 2015;261(5):939946. doi:10.1097/SLA.0000000000000747

11. Deo RC. Machine learning in medicine. Circulation. 2015;132(20):19201930. doi:10.1161/CIRCULATIONAHA.115.001593

12. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. 2019;380(14):13471358. doi:10.1056/NEJMra1814259

13. Eaton JE, Vesterhus M, McCauley BM, et al. Primary sclerosing cholangitis risk estimate tool (PREsTo) predicts outcomes of the disease: a derivation and validation study using machine learning. Hepatology. 2020;71(1):214224. doi:10.1002/hep.30085

14. Nathan H, Hyder O, Mayo SC, et al. Surgical therapy for early hepatocellular carcinoma in the modern era: a 10-year SEER-medicare analysis. Ann Surg. 2013;258(6):10221027. doi:10.1097/SLA.0b013e31827da749

15. Fritz AG. International Classification of Diseases for Oncology: ICD-O. 3. Geneva, Switzerland: World Health Organization; 2000.

16. Friedman JH. Greedy function approximation: a gradient boosting machine. Ann Stat. 2001;29(5):11891232. doi:10.1214/aos/1013203451

17. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making. 2006;26(6):565574. doi:10.1177/0272989X06295361

18. Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004;10(21):72527259. doi:10.1158/1078-0432.CCR-04-0713

19. Altekruse SF, Henley SJ, Cucinelli JE, McGlynn KA. Changing hepatocellular carcinoma incidence and liver cancer mortality rates in the United States. Am J Gastroenterol. 2014;109(4):542553. doi:10.1038/ajg.2014.11

20. Dasari BV, Kamarajah SK, Hodson J, et al. Development and validation of a risk score to predict the overall survival following surgical resection of hepatocellular carcinoma in non-cirrhotic liver. HPB (Oxford). 2020;22(3):383390. doi:10.1016/j.hpb.2019.07.007

21. Ferrer-Fbrega J, Forner A, Liccioni A, et al. Prospective validation of ab initio liver transplantation in hepatocellular carcinoma upon detection of risk factors for recurrence after resection. Hepatology. 2016;63(3):839849. doi:10.1002/hep.28339

22. Qiu J, Peng B, Tang Y, et al. CpG methylation signature predicts recurrence in early-stage hepatocellular carcinoma: results from a Multicenter Study. J Clin Oncol. 2017;35(7):734742. doi:10.1200/JCO.2016.68.2153

23. Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ. 2009;338:b2393. doi:10.1136/bmj.b2393

24. Loftus TJ, Tighe PJ, Filiberto AC, et al. Artificial intelligence and surgical decision-making. JAMA Surg. 2020;155(2):148158. doi:10.1001/jamasurg.2019.4917

25. Shindoh J, Andreou A, Aloia TA, et al. Microvascular invasion does not predict long-term survival in hepatocellular carcinoma up to 2 cm: reappraisal of the staging system for solitary tumors. Ann Surg Oncol. 2013;20(4):12231229. doi:10.1245/s10434-012-2739-y

26. Bibault JE, Chang DT, Xing L. Development and validation of a model to predict survival in colorectal cancer using a gradient-boosted machine. Gut. 2021;70(5):884889. doi:10.1136/gutjnl-2020-321799

27. Maros ME, Capper D, Jones DTW, et al. Machine learning workflows to estimate class probabilities for precision cancer diagnostics on DNA methylation microarray data. Nat Protoc. 2020;15(2):479512. doi:10.1038/s41596-019-0251-6

28. Jeong CW, Washington SL 3rd, Herlemann A, Gomez SL, Carroll PR, Cooperberg MR. The new surveillance, epidemiology, and end results prostate with watchful waiting database: opportunities and limitations. Eur Urol. 2020;78(3):335344. doi:10.1016/j.eururo.2020.01.009

View post:
Machine learning to improve prognosis prediction of EHCC | JHC - Dove Medical Press