Forecasting of Lung Cancer Incident Cases at the Small-Area Level in Victoria, Australia

Predicting lung cancer cases at the small-area level is helpful to quantify the lung cancer burden for health planning purposes at the local geographic level. Using Victorian Cancer Registry (2001-2018) data, this study aims to forecast lung cancer counts at the local government area (LGA) level over the next ten years (2019-2028) in Victoria, Australia. We used the Age-Period-Cohort approach to estimate the annual age-specific incidence and utilised Bayesian spatio-temporal models that account for non-linear temporal trends and area-level risk factors. Compared to 2001, lung cancer incidence increased by 28.82% from 1353 to 1743 cases for men and 78.79% from 759 to 1357 cases for women in 2018. Lung cancer counts are expected to reach 2515 cases for men and 1909 cases for women in 2028, with a corresponding 44% and 41% increase. The majority of LGAs are projected to have an increasing trend for both men and women by 2028. Unexplained area-level spatial variation substantially reduced after adjusting for the elderly population in the model. Male and female lung cancer cases are projected to rise at the state level and in each LGA in the next ten years. Population growth and an ageing population largely contributed to this rise.

Baidu Index and COVID-19 Epidemic Forecast: Evidence From China

With the global spread of the Coronavirus epidemic, search engine data can be a practical tool for decision-makers to understand the epidemic's trends. This article uses trend analysis data from the Baidu search engine, the most widely used in China, to analyze the public's attention to the epidemic and the demand for N95 masks and other anti-epidemic materials and information. This kind of analysis has become an important part of information epidemiology. We have analyzed the use of the keywords "Coronavirus epidemic," "N95 mask," and "Wuhan epidemic" to judge whether the introduction of real-time search data has improved the efficiency of the Coronavirus epidemic prediction model. In general, the introduction of the Baidu index, whether in-sample or out-of-sample, significantly improves the prediction efficiency of the model.

Machine Learning-Based Prediction of Growth in Confirmed COVID-19 Infection Cases in 114 Countries Using Metrics of Nonpharmaceutical Interventions and Cultural Dimensions: Model Development and Validation

BACKGROUND

National governments worldwide have implemented nonpharmaceutical interventions to control the COVID-19 pandemic and mitigate its effects.

OBJECTIVE

The aim of this study was to investigate the prediction of future daily national confirmed COVID-19 infection growth-the percentage change in total cumulative cases-across 14 days for 114 countries using nonpharmaceutical intervention metrics and cultural dimension metrics, which are indicative of specific national sociocultural norms.

METHODS

We combined the Oxford COVID-19 Government Response Tracker data set, Hofstede cultural dimensions, and daily reported COVID-19 infection case numbers to train and evaluate five non-time series machine learning models in predicting confirmed infection growth. We used three validation methods-in-distribution, out-of-distribution, and country-based cross-validation-for the evaluation, each of which was applicable to a different use case of the models.

RESULTS

Our results demonstrate high R2 values between the labels and predictions for the in-distribution method (0.959) and moderate R2 values for the out-of-distribution and country-based cross-validation methods (0.513 and 0.574, respectively) using random forest and adaptive boosting (AdaBoost) regression. Although these models may be used to predict confirmed infection growth, the differing accuracies obtained from the three tasks suggest a strong influence of the use case.

CONCLUSIONS

This work provides new considerations in using machine learning techniques with nonpharmaceutical interventions and cultural dimensions as metrics to predict the national growth of confirmed COVID-19 infections.

Trend Analysis and Forecasting the Spread of COVID-19 Pandemic in Ethiopia Using Box-Jenkins Modeling Procedure

INTRODUCTION

COVID-19, which causes severe acute respiratory syndrome, is spreading rapidly across the world, and the severity of this pandemic is rising in Ethiopia. The main objective of the study was to analyze the trend and forecast the spread of COVID-19 and to develop an appropriate statistical forecast model.

METHODOLOGY

Data on the daily spread between 13 March, 2020 and 31 August 2020 were collected for the development of the autoregressive integrated moving average (ARIMA) model. Stationarity testing, parameter testing and model diagnosis were performed. In addition, candidate models were obtained using autocorrelation function (ACF) and partial autocorrelation functions (PACF). Finally, the fitting, selection and prediction accuracy of the ARIMA models was evaluated using the RMSE and MAPE model selection criteria.

RESULTS

A total of 51,910 confirmed COVID-19 cases were reported from 13 March to 31 August 2020. The total recovered and death rates as of 31 August 2020 were 37.2% and 1.57%, respectively, with a high level of increase after the mid of August, 2020. In this study, ARIMA (0, 1, 5) and ARIMA (2, 1, 3) were finally confirmed as the optimal model for confirmed and recovered COVID-19 cases, respectively, based on lowest RMSE, MAPE and BIC values. The ARIMA model was also used to identify the COVID-19 trend and showed an increasing pattern on a daily basis in the number of confirmed and recovered cases. In addition, the 60-day forecast showed a steep upward trend in confirmed cases and recovered cases of COVID-19 in Ethiopia.

CONCLUSION

Forecasts show that confirmed and recovered COVID-19 cases in Ethiopia will increase on a daily basis for the next 60 days. The findings can be used as a decision-making tool to implement health interventions and reduce the spread of COVID-19 infection.

Impact of COVID-19 Testing Strategies and Lockdowns on Disease Management Across Europe, South America, and the United States: Analysis Using Skew-Normal Distributions

Background

As COVID-19 infections worldwide exceed 6 million confirmed cases, the data reveal that the first wave of the outbreak is coming to an end in many European countries. There is variation in the testing strategies (eg, massive testing vs testing only those displaying symptoms) and the strictness of lockdowns imposed by countries around the world. For example, Brazil's mitigation measures lie between the strict lockdowns imposed by many European countries and the more liberal approach taken by Sweden. This can influence COVID-19 metrics (eg, total deaths, confirmed cases) in unexpected ways.

Objective

This study aimed to evaluate the effectiveness of local authorities' strategies in managing the COVID-19 pandemic in Europe, South America, and the United States.

Methods

The early stage of the COVID-19 outbreak in Brazil was compared to Europe using the weekly transmission rate. Using the European data as a basis for our analysis, we examined the spread of COVID-19 and modeled curves pertaining to daily confirmed cases and deaths per million using skew-normal probability density functions. For Sweden, the United Kingdom, and the United States, we forecasted the end of the pandemic, and for Brazil, we predicted the peak value for daily deaths per million. We also discussed additional factors that could play an important role in the fight against COVID-19, such as the fast response of local authorities, testing strategies, number of beds in the intensive care unit, and isolation strategies adopted.

Results

The European data analysis demonstrated that the transmission rate of COVID-19 increased similarly for all countries in the initial stage of the pandemic but changed as the total confirmed cases per million in each country grew. This was caused by the variation in timely action by local authorities in adopting isolation measures and/or massive testing strategies. The behavior of daily confirmed cases for the United States and Brazil during the early stage of the outbreak was similar to that of Italy and Sweden, respectively. For daily deaths per million, transmission in the United States was similar to that of Switzerland, whereas for Brazil, it was greater than the counts for Portugal, Germany, and Austria (which had, in terms of total deaths per million, the best results in Europe) but lower than other European countries.

Conclusions

The fitting skew parameters used to model the curves for daily confirmed cases per million and daily deaths per million allow for a more realistic prediction of the end of the pandemic and permit us to compare the mitigation measures adopted by local authorities by analyzing their respective skew-normal parameters. The massive testing strategy adopted in the early stage of the pandemic by German authorities made a positive difference compared to other countries like Italy where an effective testing strategy was adopted too late. This explains why, despite a strictly indiscriminate lockdown, Italy's mortality rate was one of the highest in the world.

Brain Activity Foreshadows Stock Price Dynamics

Successful investing is challenging since stock prices are difficult to consistently forecast. Recent neuroimaging evidence suggests, however, that activity in brain regions associated with anticipatory affect may not only predict individual choice, but also forecast aggregate behavior out-of-sample. Thus, in two experiments, we specifically tested whether anticipatory affective brain activity in healthy humans could forecast aggregate changes in stock prices. Using functional magnetic resonance imaging, we found in a first experiment (n = 34, 6 females; 140 trials/subject) that nucleus accumbens activity forecast stock price direction, whereas anterior insula (AIns) activity forecast stock price inflections. In a second preregistered replication experiment (n = 39, 7 females) that included different subjects and stocks, AIns activity still forecast stock price inflections. Importantly, AIns activity forecast stock price movement even when choice behavior and conventional stock indicators did not (e.g., previous stock price movements), and classifier analysis indicated that forecasts based on brain activity should generalize to other markets. By demonstrating that AIns activity might serve as a leading indicator of stock price inflections, these findings imply that neural activity associated with anticipatory affect may extend to forecasting aggregate choice in dynamic and competitive environments such as stock markets.SIGNIFICANCE STATEMENT Many try but fail to consistently forecast changes in stock prices. New evidence, however, suggests that anticipatory affective brain activity may not only predict individual choice, but also may forecast aggregate choice. Assuming that stock prices index collective choice, we tested whether brain activity sampled during the assessment of stock prices could forecast subsequent changes in the prices of those stocks. In two neuroimaging experiments, a combination of previous stock price movements and brain activity in a region implicated in processing uncertainty and arousal forecast next-day stock price changes-even when behavior did not. These findings challenge traditional assumptions of market efficiency by implying that neuroimaging data might reveal "hidden information" capable of foreshadowing stock price dynamics.

Modeling and Simulation of Biogas Production in Full Scale with Time Series Analysis

Future biogas plants must be able to produce biogas according to demand, which requires proactive feeding management. Therefore, the simulation of biogas production depending on the substrate supply is assumed. Most simulation models are based on the complex Anaerobic Digestion Model No. 1 (ADM1). The ADM1 includes a large number of parameters for all biochemical and physicochemical process steps, which have to be carefully adjusted to represent the conditions of a respective full-scale biogas plant. Due to a deficiency of reliable measurement technology and process monitoring, nearly none of these parameters are available for full-scale plants. The present research investigation shows a simulation model, which is based on the principle of time series analysis and uses only historical data of biogas formation and solid substrate supply, without differentiation of individual substrates. The results of an extensive evaluation of the model over 366 simulations with 48-h horizon show a mean absolute percentage error (MAPE) of 14–18%. The evaluation is based on two different digesters and demonstrated that the model is self-learning and automatically adaptable to the respective application, independent of the substrate’s composition.

Forecasting Risk of Crop Disease with Anomaly Detection Algorithms

Information from crop disease surveillance programs and outbreak investigations provides real-world data about the drivers of epidemics. In many cases, however, only information on outbreaks is collected and data from surrounding healthy crops are omitted. Use of such data to develop models that can forecast risk/no risk of disease is therefore problematic, as information relating to the no-risk status of healthy crops is missing. This study explored a novel application of anomaly detection techniques to derive models for forecasting risk of crop disease from data composed of outbreaks only. This was done in two steps. In the training phase, the algorithms were used to learn the envelope of weather conditions most associated with historic crop disease outbreaks. In the testing phase, the algorithms were used for hindcasting of historic outbreak events. Five different anomaly detection algorithms were compared according to their accuracy in forecasting outbreaks: robust covariance, one-class k-means, Gaussian mixture model, kernel density estimation, and one-class support vector machine. A case study of potato late blight survey data from across Great Britain was used for proof of concept. The results showed that Gaussian mixture model had the highest forecast accuracy at 97.0%, followed by one-class k-means at 96.9%. There was added value in combining the algorithms in an ensemble to provide a more accurate and robust forecasting tool that can be tailored to produce region-specific alerts. The techniques used here can easily be applied to outbreak data from other crop pathosystems to derive tools for agricultural decision support.

Cycles of self-reported seizure likelihood correspond to yield of diagnostic epilepsy monitoring

OBJECTIVE

Video-electroencephalography (vEEG) is an important component of epilepsy diagnosis and management. Nevertheless, inpatient vEEG monitoring fails to capture seizures in up to one third of patients. We hypothesized that personalized seizure forecasts could be used to optimize the timing of vEEG.

METHODS

We used a database of ambulatory vEEG studies to select a cohort with linked electronic seizure diaries of more than 20 reported seizures over at least 8 weeks. The total cohort included 48 participants. Diary seizure times were used to detect individuals' multiday seizure cycles and estimate times of high seizure risk. We compared whether estimated seizure risk was significantly different between conclusive and inconclusive vEEGs, and between vEEG with and without recorded epileptic activity. vEEGs were conducted prior to self-reported seizures; hence, the study aimed to provide a retrospective proof of concept that cycles of seizure risk were correlated with vEEG outcomes.

RESULTS

Estimated seizure risk was significantly higher for conclusive vEEGs and vEEGs with epileptic activity. Across all cycle strengths, the average time in high risk during vEEG was 29.1% compared with 14% for the conclusive/inconclusive groups and 32% compared to 18% for the epileptic activity/no epileptic activity groups. On average, 62.5% of the cohort showed increased time in high risk during their previous vEEG when epileptic activity was recorded (compared to 28% of the cohort where epileptic activity was not recorded). For conclusive vEEGs, 50% of the cohort had increased time in high risk, compared to 21.5% for inconclusive vEEGs.

SIGNIFICANCE

Although retrospective, this study provides a proof of principle that scheduling monitoring times based on personalized seizure risk forecasts can improve the yield of vEEG. Forecasts can be developed at low cost from mobile seizure diaries. A simple scheduling tool to improve diagnostic outcomes may reduce cost and risks associated with delayed or missed diagnosis in epilepsy.

Analysis of Correlation between Climate Change and Human Health Based on a Machine Learning Approach

Climate change increasingly affects every aspect of human life. Recent studies report a close correlation with human health and it is estimated that global death rates will increase by 73 per 100,000 by 2100 due to changes in temperature. In this context, the present work aims to study the correlation between climate change and human health, on a global scale, using artificial intelligence techniques. Starting from previous studies on a smaller scale, that represent climate change and which at the same time can be linked to human health, four factors were chosen. Four causes of mortality, strongly correlated with the environment and climatic variability, were subsequently selected. Various analyses were carried out, using neural networks and machine learning to find a correlation between mortality due to certain diseases and the leading causes of climate change. Our findings suggest that anthropogenic climate change is strongly correlated with human health; some diseases are mainly related to risk factors while others require a more significant number of variables to derive a correlation. In addition, a forecast of victims related to climate change was formulated. The predicted scenario confirms that a prevalently increasing trend in climate change factors corresponds to an increase in victims.

Health information technology foresight for Iran: A Delphi study of experts' views to inform future policymaking

BACKGROUND

Policymakers require a systematic approach when planning for information technology needs in healthcare.

OBJECTIVE

The aim of this study was to obtain experts' predictions of future health information technology (HIT) needs until 2025 for Iran in relation to the relative importance of key technologies, expected timeframe of realisation, areas that may be impacted upon and obstacles to achieving these goals.

METHOD

This article presents results from the third phase (a Delphi study) of a larger mixed-method study. Policymakers from the Iranian Ministry of Health and faculty members from different medical universities across the country who were expert in the field of HIT were invited to participate (n = 61).

RESULTS

Participants (39) completed the first-round questionnaire and 24 completed the second. The development of personal health records (n = 32, 82.0%), the development of clinical decision-making systems (n = 30, 76.9%) and the use of business intelligence for collecting and analysing clinical and financial data (n = 32, 82.0%) were predicted to occur after 2025. The healthcare areas predicted to experience the greatest impact from most HITs were facilitating patient-provider communication and improving healthcare quality. Key barriers to achieving HITs were related to weaknesses in planning and limited financial resources for most technologies.

CONCLUSION

By identifying the areas of impact and the barriers to achieving the HIT goals, more accurate planning is possible and resources can be allocated according to priorities.

Prediction on transaction amounts of China’s CBEC with improved GM (1, 1) models based on the principle of new information priority

Benefited by e-commerce activities and information technology development, cross-border e-commerce (CBEC) has experienced rapid growth and attracted much research attention. This study takes China’s CBEC as a typical research object and intends to forecast its future development trend based on an exploration of its dynamic changing rules as a whole. The data set of transaction amounts of China’s CBEC from 2008 to 2018 was used in the modeling processes of improved grey models (GM) (1,1) proposed in this study, after which forecast results on the development of China’s CBEC from 2019 to 2020 were achieved. The experimental results reveal that, introducing the principle of new information priority to the improvement of grey models indeed works when forecasting a newly-emerging and vulnerable system like CBEC. Finally, it is predicted that China’s CBEC promises to continue to grow in the near future.

Forecasting the final disease size: comparing calibrations of Bertalanffy-Pütter models

Using monthly data from the Ebola-outbreak 2013-2016 in West Africa, we compared two calibrations for data fitting, least-squares (SSE) and weighted least-squares (SWSE) with weights reciprocal to the number of new infections. To compare (in hindsight) forecasts for the final disease size (the actual value was observed at month 28 of the outbreak) we fitted Bertalanffy-Pütter growth models to truncated initial data (first 11, 12, …, 28 months). The growth curves identified the epidemic peak at month 10 and the relative errors of the forecasts (asymptotic limits) were below 10%, if 16 or more month were used; for SWSE the relative errors were smaller than for SSE. However, the calibrations differed insofar as for SWSE there were good fitting models that forecasted reasonable upper and lower bounds, while SSE was biased, as the forecasts of good fitting models systematically underestimated the final disease size. Furthermore, for SSE the normal distribution hypothesis of the fit residuals was refuted, while the similar hypothesis for SWSE was not refuted. We therefore recommend considering SWSE for epidemic forecasts.

Atrial Fibrillation in the United Kingdom: Predicting Costs of an Emerging Epidemic Recognising and Forecasting the Cost Drivers of Atrial Fibrillation-related costs

BACKGROUND

Atrial fibrillation (AF) is the most common sustained heart arrhythmia and a major preventable cause of stroke, heart failure and dementia. AF already accounts for a significant amount of National Health Service (NHS) funding, and over the coming years is highly likely to impose a growing cost on NHS budgets and the wider UK health care system. We therefore need greater understanding of the main cost drivers (e.g. hospitalisations) of this increasingly prevalent arrhythmia. Such data would help with NHS resource planning over the next decades.

METHODS

Based on prior published data, we initially calculated the cost of AF for 1995, and then again for 2000 which was calculated from a combination of contemporary and extrapolated data from that time. These data have been used as the basis for forecasting AF costs in the UK and as a share of total NHS expenditure. AF direct costs were split between cost driver categories; General Practioner (GP) consultations, GP referred OPD (Out Patient Department) visits, prescriptions and monitoring visits, primary admissions and post-discharge OPD visits. Forecast assumptions used: (i) NHS expenditure from 2020 onwards assumed to increase at annual rate of 3%/year; and (ii) the UK inflation rate to increase by 2% annually. Sensitivity modelling of 3%, 4% and 6% projected annual increase in AF prevalence amongst the population was applied.

RESULTS

The estimated direct and proportion of NHS expenditure of AF in 2020 for each of the assumed increases of 3%, 4% and 6% would be £1,435m (0.91%), £1,741m (1.11%) and £2,548m (1.62%) respectively. For 2030, the modelling would mean that the direct costs of AF and proportion of NHS expenditure would be £2,351m (1.11%), £3,141m (1.48%), £5,562m (2.63%), respectively. For 2040, the modelling shows that the direct costs of AF and proportion of NHS expenditure would be £3,851m (1.35%), £5,668m (1.99%), £12,143m (4.27%), respectively.By far the largest contributor to the total direct AF costs in 2020 was for Primary Admissions (nearly 60%), with a further 7% with post-discharge Outpatient Clinic visits. Taken together the total for these two categories in 2020 would cost the NHS between £949m and £1,685m, depending on the projected increase in annual rate of AF prevalence.The full cost of AF related hospitalisations may be underestimated, due to the other admissions associated with a secondary coding of AF, which in 2020 are forecast to cost between £2,269m and £4,030m, depending on the annual population increase of AF prevalence. There will be an increasing number of patients discharged to a nursing home after a hospital admission associated with a principal AF diagnosis, with cost estimates for this in 2000 being £111m, and predicted to rise to between £346m and £614m by 2020.

CONCLUSIONS

Focussing on 2020, AF is predicted to directly cost the NHS a total of a minimum of £1,435m and a maximum of £2,548m (depending on AF prevalence); hence, between 0.9-1.6% of NHS expenditure, mostly from primary admissions. The total direct costs of AF would increase to 1.35-4.27% of NHS expenditure, over the next 2 decades. If hospitalisations can be avoided or reduced, we would substantially reduce the healthcare costs of AF to the NHS.

COVID-19 Symptom-Related Google Searches and Local COVID-19 Incidence in Spain: Correlational Study

BACKGROUND

COVID-19 is one of the biggest pandemics in human history, along with other disease pandemics, such as the H1N1 influenza A, bubonic plague, and smallpox pandemics. This study is a small contribution that tries to find contrasted formulas to alleviate global suffering and guarantee a more manageable future.

OBJECTIVE

In this study, a statistical approach was proposed to study the correlation between the incidence of COVID-19 in Spain and search data provided by Google Trends.

METHODS

We assessed the linear correlation between Google Trends search data and the data provided by the National Center of Epidemiology in Spain-which is dependent on the Instituto de Salud Carlos III-regarding the number of COVID-19 cases reported with a certain time lag. These data enabled the identification of anticipatory patterns.

RESULTS

In response to the ongoing outbreak, our results demonstrate that by using our correlation test, the evolution of the COVID-19 pandemic can be predicted in Spain up to 11 days in advance.

CONCLUSIONS

During the epidemic, Google Trends offers the possibility to preempt health care decisions in real time by tracking people's concerns through their search patterns. This can be of great help given the critical, if not dramatic need for complementary monitoring approaches that work on a population level and inform public health decisions in real time. This study of Google search patterns, which was motivated by the fears of individuals in the face of a pandemic, can be useful in anticipating the development of the pandemic.

COVID-19 Infection Process in Italy and Spain: Are Data Talking? Evidence From ARMA and Vector Autoregression Models

COVID-19 (coronavirus disease 2019) has spread successfully worldwide in a matter of weeks. After the example of China, all the affected countries are taking hard-confinement measures to control the infection and to gain some time to reduce the significant amount of cases that arrive at the hospital. Although the measures in China reduced the percentages of new cases, this is not seen in other countries that have taken similar measures, such as Italy and Spain. After the first weeks, the worry was whether or not the healthcare system would collapse rather than its response to the patient's needs who are infected and require hospitalization. Using China as a mirror of what could happen in our countries and with the data available, we calculated a model that forecasts the peak of the curve of infection, hospitalization, and ICU bed numbers. We aimed to review the patterns of spread of the virus in the two countries and their regions, looking for similarities that reflect the existence of a typical path in this expansive virulence and the effects of the intervention of the authorities with drastic isolation measures, to contain the outbreak. A model based on Autorregressive and moving average models (ARMA) methodology and including Chinese disease pattern as a proxy, predicts the contagious pattern robustly. Based on the prediction, the hospitalization and intensive care unit (ICU) requirements were also calculated. Results suggest a reduction in the speed of contagion during April in both countries, earlier in Spain than in Italy. The forecast advanced a significant increase in the ICU needs for Spain surpassing 8,000 units by the end of April, but for Italy, ICU needs would decrease in the same period, according to the model. We present the following predictions to inform political leaders because they have the responsibility to maintain the national health systems away from collapsing. We are confident these data could help them into decision-taking and place the capitals (from hospital beds to human resources) into the right place.

Tracking and predicting the African COVID-19 pandemic

The ongoing coronavirus disease 2019 (COVID-19) pandemic is heterogeneous throughout Africa and threatening millions of lives. Surveillance and short-term modeling forecasts are critical to provide timely information for decisions on control strategies. We use a model that explains the evolution of the COVID-19 pandemic over time in the entire African continent, parameterized by socioeconomic and geoeconomic variations and the lagged effects of social policy and meteorological history. We observed the effect of the human development index, containment policies, testing capacity, specific humidity, temperature and landlocked status of countries on the local within-country and external between-country transmission. One week forecasts of case numbers from the model were driven by the quality of the reported data. Seeking equitable behavioral and social interventions, balanced with coordinated country-specific strategies in infection suppression, should be a continental priority to control the COVID-19 pandemic in Africa.

The turning point and end of an expanding epidemic cannot be precisely forecast

Epidemic spread is characterized by exponentially growing dynamics, which are intrinsically unpredictable. The time at which the growth in the number of infected individuals halts and starts decreasing cannot be calculated with certainty before the turning point is actually attained; neither can the end of the epidemic after the turning point. A susceptible-infected-removed (SIR) model with confinement (SCIR) illustrates how lockdown measures inhibit infection spread only above a threshold that we calculate. The existence of that threshold has major effects in predictability: A Bayesian fit to the COVID-19 pandemic in Spain shows that a slowdown in the number of newly infected individuals during the expansion phase allows one to infer neither the precise position of the maximum nor whether the measures taken will bring the propagation to the inhibition regime. There is a short horizon for reliable prediction, followed by a dispersion of the possible trajectories that grows extremely fast. The impossibility to predict in the midterm is not due to wrong or incomplete data, since it persists in error-free, synthetically produced datasets and does not necessarily improve by using larger datasets. Our study warns against precise forecasts of the evolution of epidemics based on mean-field, effective, or phenomenological models and supports that only probabilities of different outcomes can be confidently given.

Development and Internal Validation of a Multivariable Prediction Model for Individual Episodic Migraine Attacks Based on Daily Trigger Exposures

OBJECTIVE

To develop and internally validate a multivariable predictive model for days with new-onset migraine headaches based on patient self-prediction and exposure to common trigger factors.

BACKGROUND

Accurate real-time forecasting of one's daily risk of migraine attack could help episodic migraine patients to target preventive medications for susceptible time periods and help decrease the burden of disease. Little is known about the predictive utility of common migraine trigger factors.

METHODS

We recruited adults with episodic migraine through online forums to participate in a 90-day prospective daily-diary cohort study conducted through a custom research application for iPhone. Every evening, participants answered questions about migraine occurrence and potential predictors including stress, sleep, caffeine and alcohol consumption, menstruation, and self-prediction. We developed and estimated multivariable multilevel logistic regression models for the risk of a new-onset migraine day vs a healthy day and internally validated the models using repeated cross-validation.

RESULTS

We had 178 participants complete the study and qualify for the primary analysis which included 1870 migraine events. We found that a decrease in caffeine consumption, higher self-predicted probability of headache, a higher level of stress, and times within 2 days of the onset of menstruation were positively associated with next-day migraine risk. The multivariable model predicted migraine risk only slightly better than chance (within-person C-statistic: 0.56, 95% CI: 0.54, 0.58).

CONCLUSIONS

In this study, episodic migraine attacks were not predictable based on self-prediction or on self-reported exposure to common trigger factors. Improvements in accuracy and breadth of data collection are needed to build clinically useful migraine prediction models.

Forecasting daily confirmed COVID-19 cases in Malaysia using ARIMA models

INTRODUCTION

The novel coronavirus infection has become a global threat affecting almost every country in the world. As a result, it has become important to understand the disease trends in order to mitigate its effects. The aim of this study is firstly to develop a prediction model for daily confirmed COVID-19 cases based on several covariates, and secondly, to select the best prediction model based on a subset of these covariates.

METHODOLOGY

This study was conducted using daily confirmed cases of COVID-19 collected from the official Ministry of Health, Malaysia (MOH) and John Hopkins University websites. An Autoregressive Integrated Moving Average (ARIMA) model was fitted to the training data of observed cases from 22 January to 31 March 2020, and subsequently validated using data on cases from 1 April to 17 April 2020. The ARIMA model satisfactorily forecasted the daily confirmed COVID-19 cases from 18 April 2020 to 1 May 2020 (the testing phase).

RESULTS

The ARIMA (0,1,0) model produced the best fit to the observed data with a Mean Absolute Percentage Error (MAPE) value of 16.01 and a Bayes Information Criteria (BIC) value of 4.170. The forecasted values showed a downward trend of COVID-19 cases until 1 May 2020. Observed cases during the forecast period were accurately predicted and were placed within the prediction intervals generated by the fitted model.

CONCLUSIONS

This study finds that ARIMA models with optimally selected covariates are useful tools for monitoring and predicting trends of COVID-19 cases in Malaysia.

Excess Patient Visits for Cough and Pulmonary Disease at a Large US Health System in the Months Prior to the COVID-19 Pandemic: Time-Series Analysis

BACKGROUND

Accurately assessing the regional activity of diseases such as COVID-19 is important in guiding public health interventions. Leveraging electronic health records (EHRs) to monitor outpatient clinical encounters may lead to the identification of emerging outbreaks.

OBJECTIVE

The aim of this study is to investigate whether excess visits where the word "cough" was present in the EHR reason for visit, and hospitalizations with acute respiratory failure were more frequent from December 2019 to February 2020 compared with the preceding 5 years.

METHODS

A retrospective observational cohort was identified from a large US health system with 3 hospitals, over 180 clinics, and 2.5 million patient encounters annually. Data from patient encounters from July 1, 2014, to February 29, 2020, were included. Seasonal autoregressive integrated moving average (SARIMA) time-series models were used to evaluate if the observed winter 2019/2020 rates were higher than the forecast 95% prediction intervals. The estimated excess number of visits and hospitalizations in winter 2019/2020 were calculated compared to previous seasons.

RESULTS

The percentage of patients presenting with an EHR reason for visit containing the word "cough" to clinics exceeded the 95% prediction interval the week of December 22, 2019, and was consistently above the 95% prediction interval all 10 weeks through the end of February 2020. Similar trends were noted for emergency department visits and hospitalizations starting December 22, 2019, where observed data exceeded the 95% prediction interval in 6 and 7 of the 10 weeks, respectively. The estimated excess over the 3-month 2019/2020 winter season, obtained by either subtracting the maximum or subtracting the average of the five previous seasons from the current season, was 1.6 or 2.0 excess visits for cough per 1000 outpatient visits, 11.0 or 19.2 excess visits for cough per 1000 emergency department visits, and 21.4 or 39.1 excess visits per 1000 hospitalizations with acute respiratory failure, respectively. The total numbers of excess cases above the 95% predicted forecast interval were 168 cases in the outpatient clinics, 56 cases for the emergency department, and 18 hospitalized with acute respiratory failure.

CONCLUSIONS

A significantly higher number of patients with respiratory complaints and diseases starting in late December 2019 and continuing through February 2020 suggests community spread of SARS-CoV-2 prior to established clinical awareness and testing capabilities. This provides a case example of how health system analytics combined with EHR data can provide powerful and agile tools for identifying when future trends in patient populations are outside of the expected ranges.

Modelling of COVID-19 Morbidity in Russia

The outbreak of COVID-19 has led to a crucial change in ordinary healthcare approaches. In comparison with emergencies re-allocation of resources for a long period of time is required and the peak utilization of the resources is also hard to predict. Furthermore, the epidemic models do not provide reliable information of the development of the pandemic's development, so it creates a high load on the healthcare systems with unforeseen duration. To predict morbidity of the novel COVID-19, we used records covering the time period from 01-03-2020 to 25-05-2020 and include sophisticated information of the morbidity in Russia. Total of 45238 patients were analyzed. The predictive model was developed as a combination of Holt and Holt-Winter models with Gradient boosting Regression. As we can see from the table 2, the models demonstrated a very good performance on the test data set. The forecast is quite reliable, however, due to the many uncertainties, only a real-world data can prove the correctness of the forecast.

The hidden Markov chain modelling of the COVID-19 spreading using Moroccan dataset

The World Health Organization (WHO) declared in March 12, 2020 the COVID-19 disease as pandemic. In Morocco, the first local transmission case was detected in March 13. The number of confirmed cases has gradually increased to reach 15,194 on July 10, 2020. To predict the COVID-19 evolution, statistical and mathematical models such as generalized logistic growth model [1], exponential model [2], segmented Poisson model [3], Susceptible-Infected-Recovered derivative models [4] and ARIMA [5] have been proposed and used. Herein, we proposed the use of the Hidden Markov Chain, which is a statistical system modelling transitions from one state (confirmed cases, recovered, active or death) to another according to a transition probability matrix to forecast the evolution of COVID-19 in Morocco from March 14, to October 5, 2020. In our knowledge the Hidden Markov Chain was not yet applied to the COVID-19 spreading. Forecasts for the cumulative number of confirmed, recovered, active and death cases can help the Moroccan authorities to set up adequate protocols for managing the post-confinement due to COVID-19. We provided both the recorded and forecasted data matrices of the cumulative number of the confirmed, recovered and active cases through the range of the studied dates.

Assessment of Countries' Preparedness and Lockdown Effectiveness in Fighting COVID-19

Objectives:

The aim of this study was to assess the risks in confronting the coronavirus disease 2019 (COVID-19) pandemic and the ongoing lockdown effectiveness in each of Italy, Germany, Spain, France, and the United States using China’s lockdown model simulation, and cases forecast until the plateau phase.

Methods:

Quantitative and qualitative historical data analysis. Total Risk Assessment (TRA) evaluation tool was used to assess the pre-pandemic stage risks, pandemic threshold fast responsiveness, and the ongoing performance until plateau. The Infected Patient Ratio (IPR) tool was developed to measure the number of patients resulting from 1 infector during the incubation period. Both IPR and TRA were used together to forecast inflection points, plateau phases, intensive care units’ and ventilators’ breakpoints, and the Total Fatality Ratio.

Results:

In Italy, Spain, France, Germany, and the United States, an inflection point is predicted within the first 15 d of April, to arrive at a plateau after another 30 to 80 d. Variations in IPR drop are expected due to variations in lockdown timing by each country, the extent of adherence to it, and the number of performed tests in each.

Conclusions:

Both qualitative (TRA) and quantitative (IPR) tools can be used together for assessing and minimizing the pandemic risks and for more precise forecasting.

Forecasting deaths of road traffic injuries in China using an artificial neural network

Objectives: This study was conducted to estimate road traffic deaths and to forecast short-term road traffic deaths in China using the Elman recurrent neural network (ERNN) model.Methods: An ERNN model was developed using reported police data of road traffic deaths in China from 2000 to 2017. Different numbers of neurons of the hidden layer were tested and different combinations of subgroup datasets have been used to develop the optimal ERNN model after normalization. The mean absolute error (MAE), the root mean square error (RMSE), and the mean absolute percentage error (MAPE) were measures of the deviation between predicted and observed values. Predicted road traffic deaths from the ERNN model and the seasonal autoregressive integrated moving average (SARIMA) model were compared using the MAPE.Results: By comparing the MAE, RMSE and MAPE of different numbers of hidden neurons and different ERNN models, the ERNN model provided the best result when the input neurons were set to 3 and hidden neurons were set to 10. The best validated neural model (3:10:1) was further applied to make predictions for the latest 12 months of deaths (MAPE = 4.83). The best SARIMA (0, 1, 1) (0, 1, 1)₁₂ model was selected from various candidate models (MAPE = 5.04). The fitted road traffic deaths using the two selected models matched closely with the observed deaths from 2000 to 2016. The ERNN models performed better than the SARIMA model in terms of prediction of 2017 deaths.Conclusions: Our results suggest that the ERNN model could be utilized to model and forecast the short-term trends accurately and to evaluate the impact of traffic safety programs when applied to historical road traffic deaths data. Forecasting traffic crash deaths will provide useful information to measure burden of road traffic injuries in China.