Decision Support Aids Research Articles

The use of artificial intelligence to aid the diagnosis of lung cancer - A retrospective-cohort study.

AI software in the form of deep learning-based automatic detection (DLAD) algorithms for chest X-ray (CXR) interpretation have shown success in early detection of lung cancer (LC), however, there remains uncertainty related to clinical validation. CXRs and their corresponding chest-CT scans were retrospectively collated from a single institution between January 2019-2020. A commercially available AI software was used to evaluate 320 CXRs (<6 years prior-to-diagnosis) from 105 positive LC patients and 103 negative controls. Clinical reports were extracted and coded to correlate against AI findings. Of 105 LC patients, (57[55%] men, median [IQR] age 73[68-83] years), clinical reports identified LC in 64 (61%) whereas AI identified LC in 95 (90%). AI diagnostic (image-level) and prognostic (patient-level) sensitivities were 57.6% and 90.0%, (81% in correct location), respectively. On CXRs performed >12 months prior to LC diagnosis, the AI detected nodules in 24(23%) cases of which 22/24 had negative clinical reports for lung nodule/mass. The potential median reduction in time-to-diagnosis for cases where AI identified nodule(s) on previous CXR, but clinical reports negative, was 193[IQR 42-598] days. Of the 103 'negative' controls (48[47%] men, median [IQR] age 69[61-77] years) 20 patients had a nodule abnormality score above the threshold, generating a false-positive rate of 19%. The AI software showed excellent performance in detecting LCs that initially went undetected on CXR. The algorithm has potential to increase LC detection rates and reduce time-to-diagnosis. Using the AI, in conjunction with a trained observer, could increase reporting accuracy and potentially improve clinical outcomes. This study demonstrated the benefits and pitfalls associated with using AI in a clinical setting. It provides further evidence for utilising decision-support aids within clinical practice.

The American Cancer Society National Lung Cancer Roundtable strategic plan: Provider engagement and outreach.

The American Cancer Society National Lung Cancer Roundtable strategic plan for provider engagement and outreach addresses barriers to the uptake of lung cancer screening, including lack of provider awareness and guideline knowledge about screening, concerns about potential harms from false-positive examinations, lack of time to implement workflows within busy primary care practices, insufficient infrastructure and administrative support to manage a screening program and patient follow-up, and implicit bias based on sex, race/ethnicity, social class, and smoking status. Strategies to facilitate screening include educational programming, clinical reminder systems within the electronic medical record, decision support aids, and tools to track nodules that can be implemented across a diversity of practices and health care organizational structures. PLAIN LANGUAGE SUMMARY: The American Cancer Society National Lung Cancer Roundtable strategic plan to reduce deaths from lung cancer includes strategies designed to support health care professionals, to better understand lung cancer screening, and to support adults who are eligible for lung cancer screening by providing counseling, referral, and follow-up.

Development of a Novel Mobile Health App to Empower Young People With Type 1 Diabetes to Exercise Safely: Co-Design Approach.

Blood glucose management around exercise is challenging for youth with type 1 diabetes (T1D). Previous research has indicated interventions including decision-support aids to better support youth to effectively contextualize blood glucose results and take appropriate action to optimize glucose levels during and after exercise. Mobile health (mHealth) apps help deliver health behavior interventions to youth with T1D, given the use of technology for glucose monitoring, insulin dosing, and carbohydrate counting. We aimed to develop a novel prototype mHealth app to support exercise management among youth with T1D, detail the application of a co-design process and design thinking principles to inform app design and development, and identify app content and functionality that youth with T1D need to meet their physical activity goals. A co-design approach with a user-centered design thinking framework was used to develop a prototype mHealth app "acT1ve" during the 18-month design process (March 2018 to September 2019). To better understand and respond to the challenges among youth with diabetes when physically active, 10 focus groups were conducted with youth aged 13-25 years with T1D and parents of youth with T1D. Thereafter, we conducted participatory design workshops with youth to identify key app features that would support individual needs when physically active. These features were incorporated into a wireframe, which was critically reviewed by participants. A beta version of "acT1ve" was built in iOS and android operating systems, which underwent critical review by end users, clinicians, researchers, experts in exercise and T1D, and app designers. Sixty youth with T1D, 14 parents, 6 researchers, and 10 clinicians were engaged in the development of "acT1ve." acT1ve included key features identified by youth, which would support their individual needs when physically active. It provided advice on carbohydrates and insulin during exercise, information on hypoglycemia treatment, pre- and postexercise advice, and an educational food guide regarding exercise management. "acT1ve" contained an exercise advisor algorithm comprising 240 pathways developed by experts in diabetes and exercise research. Based on participant input during exercise, acT1ve provided personalized insulin and carbohydrate advice for exercise lasting up to 60 minutes. It also contains other features including an activity log, which displays a complete record of the end users' activities and associated exercise advice provided by the app's algorithm for later reference, and regular reminder notifications for end users to check or monitor their glucose levels. The co-design approach and the practical application of the user-centered design thinking framework were successfully applied in developing "acT1ve." The design thinking processes allowed youth with T1D to identify app features that would support them to be physically active, and particularly enabled the delivery of individualized advice. Furthermore, app development has been described in detail to help guide others embarking on a similar project. Australian New Zealand Clinical Trials Registry ACTRN12619001414101; https://tinyurl.com/mu9jvn2d.

Using anthropomorphism, transparency, and feedback to improve phishing detection aids

Phishing is a common cybersecurity threat to email users. An automated phishing decision-support aid can help users identify suspicious emails. The aid’s success depends on both the aid’s capability and the user’s trust in and usage of the aid. In this study, 465 participants were asked to judge phishing emails with an automated decision aid. We measured how users’ trust and decision making were affected by the type of aid (human, AI, text, no aid), gender (male, female), decision transparency (reasoning for aid’s decision present, absent), as well as feedback (present, absent). We found that an aid was helpful regardless of its anthropomorphic appearance (human vs. AI) and gender. Transparency was helpful with the human aid, but not the AI aid. Feedback effectively improved trust in all the aids, although it helped aid retention only for the text aid. Participants had overall positive comments about the aid and found it helpful. The findings can be applied to automated aid design to understand potential avenues for improving users’ trust, performance, and aid use.

Dynamic decision making for situational awareness using drones: Requirements, identification and comparison of decision support methods

Decision makers increasingly operate in real-time information-rich environments, where limited time is available for interpreting data to inform decisions. These environments are driven by static or mobile sensing devices that can provide numerous dynamic data points. A prominent approach in this space is to utilise drones, which can be deployed to gather targeted information. However, deciding how best to deploy available drones is nontrivial, and stands to benefit from decision support aids that plan routes. Such a system must operate under time constraints created by the changing attributes of routes as the situation unfolds. This study describes a dynamic decision support system (DSS) for situational awareness with drones. The system applies Multi-Criteria Decision Making (MCDM) methods within a dynamic genetic algorithm to provide a continuously revised ranking of routes. Five desiderata for dynamic decision support are presented. It is shown how a dynamic DSS can be equipped with declarative specification of preferences (Desiderata 1), dynamic revision of recommendations (Desiderata 2), and high diversity of options (Desiderata 3). The study then compares four MCDM methods, namely the Weighted Product Model (WPM), the Analytic Hierarchy Process (AHP), the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), and the Preference Ranking Organization METHod for Enrichment Evaluation (PROMETHEE), with regards to how consistently they trade-off between criteria (Desiderata 4) and the stability of results under small changes to criteria values (Desiderata 5). To evaluate the trade-offs between criteria we analyse the smoothness of change in criteria outcomes as criteria weightings increase for each algorithm. The outcomes are calculated by automating the selection of routes in a case study that applies drones to the task of harbour management. The stability of results for the different MCDM methods are compared. Perturbations were applied to sets of routes ranked by each algorithm then each algorithm was reapplied and the magnitude of the changes in ranking was assessed. Overall, TOPSIS was found to be the algorithm which made the most consistent trade-offs between criteria, only under-performing another algorithm with respect to a single criterion. AHP and WPM were the next most consistent algorithms and PROMETHEE was the least consistent algorithm. TOPSIS was also found to be the most stable method under small changes to criteria values. AHP was the second most stable, followed by PROMETHEE and WPM respectively. The results show that TOPSIS achieves the best result for both Desiderata 4 and 5 and consequently the study finds TOPSIS to be an appropriate MCDM method for dynamic decision support.

Implementing genetic testing in diabetes: Knowledge, perceptions of healthcare professionals, and barriers in a developing country.

Maturity-Onset Diabetes of the Young (MODY) is an unusual type of diabetes often missed in clinical practice, especially in Africa. Treatment decisions for MODY depend on a precise diagnosis, only made by genetic testing. We aimed to determine MODY knowledge among Nigerian healthcare professionals (HCPs), their perceptions, and barriers to the implementation of genetic testing in diabetes patients. A cross-sectional survey was conducted among doctors and nurses in three levels of public and private healthcare institutions in Ibadan, Nigeria, from December 2018 to June 2019. In all, 70% and 30% of a total 415 participants were recruited from public and private centers, respectively. HCPs were recruited in a 60:40% ratio, respectively. A 51-item instrument was used to assess MODY knowledge, perceptions of HCPs, and barriers to the implementation of genetic testing in diabetes patients. In the survey, 43.4% self-rated their current MODY knowledge to be at least moderate. About 68%, 73% and 86%, respectively, correctly answered 3 of 5 questions on basic genetics' knowledge. However, only 1 of 7 MODY-specific questions was answered correctly by 72.7% of the respondents. The mean basic genetics and MODY-specific knowledge scores were 2.6/5 (SD=1.0) and 1.8/9 (SD=1.3), respectively. Multiple linear regression showed higher mean scores among those aged 30-49 years, those with degrees and fellowships (except PhD), and general practitioners; 360 (80.0%) perceived that genetic testing plays a central role in diabetes care. Barriers to genetic testing were lack of access to testing facilities, guidance on the use of and updates/educational materials on genetic testing (82.7%, 62.1% and 50.3%, respectively). The level of MODY awareness and knowledge among Nigerian HCPs is unacceptably low with a lack of access to genetic testing facilities. These can hinder the implementation of precision diabetes medicine. Increased awareness, provision of decision support aids, and genetic testing facilities are urgently needed.

Predicting Rheological Properties of Wheat Dough from Flour Properties Using NIR Coupled with Artificial Neural Network

Highlights A multi-layered perceptron type artificial neural network (ANN) was developed to predict the farinograph properties of wheat dough. ANN models with two to four hidden layers were developed for each Farinograph response, i.e., water absorption, dough development time, and dough stability. Permutation importance and Shapley values analysis emphasized the significance of protein content in model prediction. The developed model would serve as a decision support tool for flour mill and bakery managers and would assist in adapting flour mill settings and modifying processing parameters in bakeries. Abstract. Farinograph analysis serves as the standard for determining the baking quality of wheat flour. It measures the water absorption (WA), dough development time (DDT), and stability (DS) of wheat dough. These characteristics depend on wheat flour properties such as protein content, moisture, ash, and falling number. Additionally, farinograph analysis is costly, time-consuming, and requires skilled personnel. Therefore, an artificial neural network (ANN) model was developed to predict the farinograph properties of wheat dough as a function of flour properties. The models were developed using data from 192 wheat samples. Multi-layer perceptron-type feed-forward ANN models with increasing complexity were developed for each response variable, i.e., ANN-WA, ANN-DDT, and ANN-DS, and model success was evaluated via mean squared error (MSE) and correlation coefficient (r). The optimal models had two to four hidden layers, each with five to sixty neurons, and exhibited the lowest MSE and highest r values. In terms of predictive performance, the models ANN-WA and ANN-DDT (r = 0.79) demonstrated superior performance when compared with ANN-DS (r = 0.63). A feature importance analysis was conducted to provide insight on variable contributions, underscoring the significance of flour protein content in the model’s prediction. The study explored the applicability of data-driven ANN models in predicting the rheological characteristics of dough. The developed models could serve as a decision support tool and aid millers in adjusting mill settings and bakers in modifying dough mixing based on dough rheology. Keywords: Artificial neural network, Dough rheology, Farinograph analysis, Milling, NIR.

Application of Computer Aided Decision Support System Based on Data Mining in Public Management

Application of Computer Aided Decision Support System Based on Data Mining in Public Management

Randomized Controlled Trial of Clinical Guidelines Versus Interactive Decision-Support for Improving Medical Trainees' Confidence with Latent Tuberculosis Care.

In order to eliminate tuberculosis (TB) in the USA, primary care providers must take on an expanded role in the diagnosis and management of latent tuberculosis infection (LTBI). Clinical practice guidelines and recommendations exist for LTBI management, but there is a need for innovative tools to improve medical students' and residents' knowledge of evidence-based practices for LTBI testing and treatment. To assess the impact of LTBI-ASSIST, a free online decision support aid, as a novel educational tool and mechanism of delivering clinical practice guidelines for medical trainees. A single site, randomized controlled trial of trainees delivered by electronic survey. Medical students and Internal Medicine residents at the Johns Hopkins University School of Medicine. Participants were randomized in 1:1 ratio to receive the US clinical practice guidelines and recommendations for Latent TB management (control arm) or the guidelines plus an introduction to LTBI-ASSIST (LTBI-ASSIST arm) as they completed a case-based knowledge assessment and reported confidence with domains of LTBI care. (1) Proportion of questions answered correctly on a case-based knowledge assessment; (2) change in reported confidence with domains of LTBI care. One hundred and thirty participants completed the knowledge assessment. Those randomized to receive the LTBI-ASSIST Tool performed better on the case-based knowledge assessment with a mean score of 75.9% (95% CI: 70.6-81.1), compared to 57.4% (52.8-62.0) in the group that received the guidelines only (p <0.001). Similarly, the LTBI-ASSIST group reported a higher change in confidence (measured as post-assessment confidence minus pre-assessment confidence), compared to the control group, in six of the seven domains of LTBI care. LTBI-ASSIST can be an effective supplement to existing guidelines in educating medical trainees and helping providers find evidence-based, guideline-supported answers for questions encountered in clinical practice. NIH Clinical Trial Registry No. NCT05772065.

Revisiting the promise, practice and progress of resuscitative endovascular balloon occlusion of the aorta.

The use of Resuscitative Endovascular Balloon Occlusion of the Aorta (REBOA) to temporarily control bleeding and improve central perfusion in critically injured trauma patients remains a controversial topic. In the last decade, select trauma services around the world have gained experience with REBOA. We discuss the recent observational data together with the initial results of the first randomized control trial and provide a view on the next steps for REBOA in trauma resuscitation. While the observational data continue to be conflicting, the first randomized control trial signals that in the UK, in-hospital REBOA is associated with harm. Likely a result of delays to haemorrhage control, views are again split on whether to abandon complex interventions in bleeding trauma patients and to only prioritize transfer to the operating room or to push REBOA earlier into the post injury phase, recognizing that some patients will not survive without intervention. Better understanding of cardiac shock physiology provides a new lens in which to evaluate REBOA through. Patient selection remains a huge challenge. Invasive blood pressure monitoring, combined with machine learning aided decision support may assist clinicians and their patients in the future. The use of REBOA should not delay definitive haemorrhage control in those patients without impending cardiac arrest.

Development of a computer aided decision support system for the design of drip irrigation laterals

Development of a computer aided decision support system for the design of drip irrigation laterals

Research on Military Logistics Data Warehouse

According to military logistics data warehouse dealing with large-scale data effect is poor. This paper puts forward the construction scheme of dual computing engine military logistics data warehouse based on hive in order to improve the military logistics support capability. The scheme adopts Hadoop and Spark dual search architecture to support efficient storage of heterogeneous data. In addition, a data block placement strategy based on correlation analysis was used. This article use the intersection relationship matrix and the concurrency relationship matrix to calculate the correlation between the data block placed on the node and the data block to be placed. The access frequency of the target data block was integrated to select the appropriate node storage. The scheme can effectively improve hive’s query efficiency and provide more effective information aided decision support for commanders.

Identification of Dental Implant Systems Using a Large-Scale Multicenter Data Set

This study aimed to evaluate the efficacy of deep learning (DL) for the identification and classification of various types of dental implant systems (DISs) using a large-scale multicenter data set. We also compared the classification accuracy of DL and dental professionals. The data set, which was collected from 5 college dental hospitals and 10 private dental clinics, contained 37,442 (24.8%) periapical and 113,291 (75.2%) panoramic radiographic images and consisted of a total of 10 manufacturers and 25 different types of DISs. The classification accuracy of DL was evaluated using a pretrained and modified ResNet-50 architecture, and comparison of accuracy performance and reading time between DL and dental professionals was conducted using a self-reported questionnaire. When comparing the accuracy performance for classification of DISs, DL (accuracy: 82.0%; 95% confidence interval [CI], 75.9%–87.0%) outperformed most of the participants (mean accuracy: 23.5% ± 18.5%; 95% CI, 18.5%–32.3%), including dentists specialized (mean accuracy: 43.3% ± 20.4%; 95% CI, 12.7%–56.2%) and not specialized (mean accuracy: 16.8% ± 9.0%; 95% CI, 12.8%–20.9%) in implantology. In addition, DL tends to require lesser reading and classification time (4.5 min) than dentists who specialized (75.6 ± 31.0 min; 95% CI, 13.1–78.4) and did not specialize (91.3 ± 38.3 min; 95% CI, 74.1–108.6) in implantology. DL achieved reliable outcomes in the identification and classification of various types of DISs, and the classification accuracy performance of DL was significantly superior to that of specialized or nonspecialized dental professionals. DL as a decision support aid can be successfully used for the identification and classification of DISs encountered in clinical practice.

Prognostication for Patients Receiving Palliative Radiation Therapy

Estimation of patient prognosis plays a central role in guiding decision making for the palliative management of metastatic disease, and a number of statistical models have been developed to provide survival estimates for patients in this context. In this review, we discuss several well-validated survival prediction models for patients receiving palliative radiotherapy to sites outside of the brain. Key considerations include the type of statistical model, model performance measures and validation procedures, studies' source populations, time points used for prognostication, and details of model output. We then briefly discuss underutilization of these models, the role of decision support aids, and the need to incorporate patient preference in shared decision making for patients with metastatic disease who are candidates for palliative radiotherapy.

Discovering Themes in Deep Brain Stimulation Research Using Explainable Artificial Intelligence.

Deep brain stimulation is a treatment that controls symptoms by changing brain activity. The complexity of how to best treat brain dysfunction with deep brain stimulation has spawned research into artificial intelligence approaches. Machine learning is a subset of artificial intelligence that uses computers to learn patterns in data and has many healthcare applications, such as an aid in diagnosis, personalized medicine, and clinical decision support. Yet, how machine learning models make decisions is often opaque. The spirit of explainable artificial intelligence is to use machine learning models that produce interpretable solutions. Here, we use topic modeling to synthesize recent literature on explainable artificial intelligence approaches to extracting domain knowledge from machine learning models relevant to deep brain stimulation. The results show that patient classification (i.e., diagnostic models, precision medicine) is the most common problem in deep brain stimulation studies that employ explainable artificial intelligence. Other topics concern attempts to optimize stimulation strategies and the importance of explainable methods. Overall, this review supports the potential for artificial intelligence to revolutionize deep brain stimulation by personalizing stimulation protocols and adapting stimulation in real time.

A design concept of big data analytics model for managers in hospitality industries.

The hospitality and tourism sector has long played a significant role in Australia's economy, especially in regional areas. Due to the onslaught of COVID-19, numerous businesses have experienced lockdowns, restrictions, and closures due to the fact that people's activity in restaurants, shopping centers, and recreational destinations was restricted, and many other places went into hibernation. After about 2years since the outbreak, businesses in this sector are gradually starting to reopen and revitalize themselves, but in order to have better decision support about the future of this sector, thus being able to plan, businesses are suffering from an effective analytics solution due to the lack of broken data trends. Starting from fresh day-to-day real-time big data, the study aims to develop a new data analytics model, adopting the design science research methodology, which can provide invaluable options and techniques to make prediction easier from immediate past datasets. This study introduces an innovative design artifact as a big data solution for hospitality managers to utilize analytics for predictive strategic decision-making in post-COVID situation. The artifact can also be generalized for other sectors with tailoring aspects which are subject to further studies. The proposed artifact is then compared with other design artifacts related to big data solutions where it outperforms them in terms of comprehensiveness. The proposed artifact also shows promises for primarily available UGC in managers' decision support aids.

Quantifying the Implementation and Cost of a Multisite Antibiotic Stewardship Intervention for Asymptomatic Bacteriuria.

The intensity of an antibiotic stewardship intervention to achieve clinical impact is not known. We conducted a multisite dissemination project of an intervention to reduce treatment of asymptomatic bacteriuria (ASB) and studied: (1) the association between implementation metrics and clinical outcomes and (2) the cost of implementation. A central site facilitated a multimodality intervention to decrease unnecessary urine cultures and antibiotic treatment in patients with ASB at 4 Veterans Affairs medical centers. The intervention consisted of a decision support aid algorithm and interactive teaching cases that provided in the moment audit and feedback on how to manage ASB. Implementation outcomes included minutes spent in intervention delivery, number of healthcare professionals reached, and number of sessions delivered. Clinical outcomes included days of antibiotic therapy (DOT), length of antibiotic therapy (LOT), and number of urine cultures ordered per 1000 bed days. Personnel reported weekly time logs. Minutes spent in intervention delivery were inversely correlated with two clinical outcomes, DOT (R -0.3, P = .04) and LOT (R -0.3, P = .02). Number of healthcare professionals reached and number of sessions delivered were not correlated with clinical outcomes of DOT (R -0.003, P = .98, R = -0.059, P = .69) or LOT (R +0.073, P = .62, R -0.102, P = .49). Physician champions spent an average of 3.8% of effort on the intervention. The implementation cost was USD 22,299/year per site on average. The amount of time local teams spent in delivery of an antibiotic stewardship intervention was correlated with the desired decrease in antibiotic use. Implementing this successful antibiotic stewardship intervention required minimal time.

Improving automated thyroid cancer classification of frozen sections by the aid of virtual image translation and stain normalization

Frozen sections are rapidly generated during surgical interventions. This allows surgeons to wait for histological findings during the interventions in order to base intra-surgical decisions on the outcome of the histology. However, compared to paraffin sections the quality of frozen sections is often strongly reduced, leading to a lower diagnostic accuracy. Deep learning-based image translation technology facilitates a virtual conversion between different native imaging technologies with the potential of translating a frozen section into a virtual paraffin section. Stain normalization can be applied to adjust further unequal image characteristics. We investigated the effect of deep learning-based image translation, conventional image normalization and a combination of these techniques on computer aided decision support systems for thyroid cancer diagnostics. For classification, a bag-of-words approach, based on convolutional neural network features, k-means clustering and a support vector machine were employed. While stain normalization led to a decreased overall classification accuracy (0.703 vs 0.727), image translation led to an increased mean score (0.770). A combination of both, image translation and normalization increased the accuracy even further (0.844) and clearly reduced the gap to the post-operative paraffin sections (0.902). Deep learning-based image translation proved to be a powerful tool to enhance accuracy of computer aided diagnosis which clearly outperformed conventional stain translation. This work provides a strong motivation for performing a study with expert pathologists performing the categorization of frozen sections and the corresponding improved sections, to investigate whether a similar effect is achieved in a clinical setting.

Inventory policy determination in MSMEs using intuitionistic fuzzy sets based on learning aided decision support system

Inventory policy determination in MSMEs using intuitionistic fuzzy sets based on learning aided decision support system

The Impact of Cultural Embeddedness on the Implementation of an Artificial Intelligence Program at Triage: A Qualitative Study.

Triage requires rapid determination of acuity and resources. Current modalities allow for individual judgment, with varied application of algorithmic rules. Although artificial intelligence can improve triage accuracy, gaps remain in understanding implementation facilitators and barriers, especially those related to the cultural understandings by nurses of emergency department presentations. The purpose of this study was to explore the cultural and technological elements of the implementation of an artificial intelligence clinical decision support aid (i.e., KATE) in an emergency nursing triage process in an urban community hospital on the West Coast of the United States. An exploratory qualitative study using semi-structured small group and individual interviews and constant comparison analysis strategies. The sample comprised 13 emergency department triage nurses at one site. Campinha-Bacote's theory of cultural competence framed the study. Responses yielded the overall theme of We know these people and we know these things. Supporting categories included the problem of aire; just another checkbox; gut trumps data; higher acuity with no resources; and technology as a safety net. Participants reported reliance on clinical experience and cultural knowledge to assign acuity. The implementation of an artificial intelligence program was initially received skeptically due to the acontextual nature of AI, but grew to be perceived as a safety net for triage decision making among emergency nurses.