Within this paper we describe the background context and discussions that lead to the selection of the theme for the NI2006 Post Congress Conference. The rationale for the choice of the overall theme is introduced, together with the selection of the five main sub-themes, on which the group discussions were based, and that lead to the production of the group papers that form the core of these Proceedings.

Studies in Health Technology and Informatics

17th International Conference on Medical Image Computing and Computer-Assisted Intervention

The practicability of deep learning techniques has been demonstrated by their successful implementation in varied fields, including diagnostic imaging for clinicians. In accordance with the increasing demands in the healthcare industry, techniques for automatic prediction and detection are being widely researched. Particularly in dentistry, for various reasons, automated mandibular canal detection has become highly desirable. The positioning of the inferior alveolar nerve (IAN), which is one of the major structures in the mandible, is crucial to prevent nerve injury during surgical procedures. However, automatic segmentation using Cone beam computed tomography (CBCT) poses certain difficulties, such as the complex appearance of the human skull, limited number of datasets, unclear edges, and noisy images. Using work-in-progress automation software, experiments were conducted with models based on 2D SegNet, 2D and 3D U-Nets as preliminary research for a dental segmentation automation tool. The 2D U-Net with adjacent images demonstrates higher global accuracy of 0.82 than naive U-Net variants. The 2D SegNet showed the second highest global accuracy of 0.96, and the 3D U-Net showed the best global accuracy of 0.99. The automated canal detection system through deep learning will contribute significantly to efficient treatment planning and to reducing patients' discomfort by a dentist. This study will be a preliminary report and an opportunity to explore the application of deep learning to other dental fields.

/pdf/automatic-mandibular-canal-detection-using-a-deep-ajhmdv9spr.pdf

Automatic mandibular canal detection using a deep convolutional neural network

Eating disorders, such as anorexia nervosa are a major health concern affecting many young individuals. Given the extensive adoption of social media technologies in the anorexia affected demographic, we study behavioral characteristics of this population focusing on the social media Tumblr. Aligned with observations in prior literature, we find the presence of two prominent anorexia related communities on Tumblr -- pro-anorexia and pro-recovery. Empirical analyses on several thousand Tumblr posts show use of the site as a media-rich platform replete with triggering content for enacting anorexia as a lifestyle choice. Through use of common pro-anorexia tags, the pro-recovery community however attempts to "permeate" into the pro-anorexia community to educate them of the health risks of anorexia. Further, the communities exhibit distinctive affective, social, cognitive, and linguistic style markers. Compared with recover- ing anorexics, pro-anorexics express greater negative affect, higher cognitive impairment, and greater feelings of social isolation and self-harm. We also observe that these characteristics may be used in a predictive setting to detect anorexia content with 80% accuracy. Based on our findings, clinical implications of detecting anorexia related content on social media are discussed.

Anorexia on Tumblr: A Characterization Study.

The aim of this study was to evaluate the success of the artificial intelligence (AI) system in implant planning using three-dimensional cone-beam computed tomography (CBCT) images. Seventy-five CBCT images were included in this study. In these images, bone height and thickness in 508 regions where implants were required were measured by a human observer with manual assessment method using InvivoDental 6.0 (Anatomage Inc. San Jose, CA, USA). Also, canals/sinuses/fossae associated with alveolar bones and missing tooth regions were detected. Following, all evaluations were repeated using the deep convolutional neural network (Diagnocat, Inc., San Francisco, USA) The jaws were separated as mandible/maxilla and each jaw was grouped as anterior/premolar/molar teeth region. The data obtained from manual assessment and AI methods were compared using Bland–Altman analysis and Wilcoxon signed rank test. In the bone height measurements, there were no statistically significant differences between AI and manual measurements in the premolar region of mandible and the premolar and molar regions of the maxilla (p > 0.05). In the bone thickness measurements, there were statistically significant differences between AI and manual measurements in all regions of maxilla and mandible (p < 0.001). Also, the percentage of right detection was 72.2% for canals, 66.4% for sinuses/fossae and 95.3% for missing tooth regions. Development of AI systems and their using in future for implant planning will both facilitate the work of physicians and will be a support mechanism in implantology practice to physicians.

/pdf/a-deep-learning-approach-for-dental-implant-planning-in-cone-1qk5cfvoib.pdf

A deep learning approach for dental implant planning in cone-beam computed tomography images.

Machine learning and deep learning have provided us with an exploration of a whole new research era. As more data and better computational power become available, they have been implemented in various fields. The demand for artificial intelligence in the field of health informatics is also increasing and we can expect to see the potential benefits of artificial intelligence applications in healthcare. Deep learning can help clinicians diagnose disease, identify cancer sites, identify drug effects for each patient, understand the relationship between genotypes and phenotypes, explore new phenotypes, and predict infectious disease outbreaks with high accuracy. In contrast to traditional models, its approach does not require domain-specific data pre-process, and it is expected that it will ultimately change human life a lot in the future. Despite its notable advantages, there are some challenges on data (high dimensionality, heterogeneity, time dependency, sparsity, irregularity, lack of label) and model (reliability, interpretability, feasibility, security, scalability) for practical use. This article presents a comprehensive review of research applying deep learning in health informatics with a focus on the last five years in the fields of medical imaging, electronic health records, genomics, sensing, and online communication health, as well as challenges and promising directions for future research. We highlight ongoing popular approaches' research and identify several challenges in building deep learning models.

DeepHealth: Deep Learning for Health Informatics

Early and accurate prediction of the need for intubation may provide more time for preparation and increase safety margins by avoiding high risk late intubation. This study evaluates whether machine learning can predict the need for intubation within 24 h using commonly available bedside and laboratory parameters taken at critical care admission. We extracted data from 2 large critical care databases (MIMIC-III and eICU-CRD). Missing variables were imputed using autoencoder. Machine learning classifiers using logistic regression and random forest were trained using 60% of the data and tested using the remaining 40% of the data. We compared the performance of logistic regression and random forest models to predict intubation in critically ill patients. After excluding patients with limitations of therapy and missing data, we included 17,616 critically ill patients in this retrospective cohort. Within 24 h of admission, 2,292 patients required intubation, whilst 15,324 patients were not intubated. Blood gas parameters (PaO2, PaCO2, HCO3-), Glasgow Coma Score, respiratory variables (respiratory rate, SpO2), temperature, age, and oxygen therapy were used to predict intubation. Random forest had AUC 0.86 (95% CI 0.85-0.87) and logistic regression had AUC 0.77 (95% CI 0.76-0.78) for intubation prediction performance. Random forest model had sensitivity of 0.88 (95% CI 0.86-0.90) and specificity of 0.66 (95% CI 0.63-0.69), with good calibration throughout the range of intubation risks. The results showed that machine learning could predict the need for intubation in critically ill patients using commonly collected bedside clinical parameters and laboratory results. It may be used in real-time to help clinicians predict the need for intubation within 24 h of intensive care unit admission.

/pdf/predicting-the-need-for-intubation-in-the-first-24-h-after-3b1evd66oj.pdf

Predicting the need for intubation in the first 24 h after critical care admission using machine learning approaches

BACKGROUND: Unprecedented public health measures have been used during this coronavirus 2019 (COVID-19) pandemic to control the spread of SARS-CoV-2 virus. It is a challenge to implement timely and appropriate public health interventions. METHODS AND FINDINGS: Population and COVID-19 epidemiological data between 21st January 2020 to 15th November 2020 from 216 countries and territories were included with the implemented public health interventions. We used deep reinforcement learning, and the algorithm was trained to enable agents to try to find optimal public health strategies that maximized total reward on controlling the spread of COVID-19. The results suggested by the algorithm were analyzed against the actual timing and intensity of lockdown and travel restrictions. Early implementations of the actual lockdown and travel restriction policies, usually at the time of local index case were associated with less burden of COVID-19. In contrast, our agent suggested to initiate at least minimal intensity of lockdown or travel restriction even before or on the day of the index case in each country and territory. In addition, the agent mostly recommended a combination of lockdown and travel restrictions and higher intensity policies than the policies implemented by governments, but did not always encourage rapid full lockdown and full border closures. The limitation of this study was that it was done with incomplete data due to the emerging COVID-19 epidemic, inconsistent testing and reporting. In addition, our research focuses only on population health benefits by controlling the spread of COVID-19 without balancing the negative impacts of economic and social consequences. INTERPRETATION: Compared to actual government implementation, our algorithm mostly recommended earlier intensity of lockdown and travel restrictions. Reinforcement learning may be used as a decision support tool for implementation of public health interventions during COVID-19 and future pandemics.

/pdf/deep-reinforcement-learning-approaches-for-global-public-4il8yjlom4.pdf

Deep reinforcement learning approaches for global public health strategies for COVID-19 pandemic.

Artificial intelligence has provided us with an exploration of a whole new research era. As more data and better computational power become available, the approach is being implemented in various fields. The demand for it in health informatics is also increasing, and we can expect to see the potential benefits of its applications in healthcare. It can help clinicians diagnose disease, identify drug effects for each patient, understand the relationship between genotypes and phenotypes, explore new phenotypes or treatment recommendations, and predict infectious disease outbreaks with high accuracy. In contrast to traditional models, recent artificial intelligence approaches do not require domain-specific data pre-processing, and it is expected that it will ultimately change life in the future. Despite its notable advantages, there are some key challenges on data (high dimensionality, heterogeneity, time dependency, sparsity, irregularity, lack of label, bias) and model (reliability, interpretability, feasibility, security, scalability) for practical use. This article presents a comprehensive review of research applying artificial intelligence in health informatics, focusing on the last seven years in the fields of medical imaging, electronic health records, genomics, sensing, and online communication health, as well as challenges and promising directions for future research. We highlight ongoing popular approaches' research and identify several challenges in building models.

Gloria Hyunjung Kwak

Papers

Automatic mandibular canal detection using a deep convolutional neural network

DeepHealth: Deep Learning for Health Informatics

Predicting the need for intubation in the first 24 h after critical care admission using machine learning approaches

Deep reinforcement learning approaches for global public health strategies for COVID-19 pandemic.

DeepHealth: Review and challenges of artificial intelligence in health informatics