There is a general consensus among Earth scientists that melting of land ice greatly contributes to sea-level rise (SLR) and that future warming will exacerbate the risks posed to human civilization. As land ice is lost to the oceans, both the Earth’s gravitational and rotational potentials are perturbed, resulting in strong spatial patterns in SLR, termed sea-level fingerprints. We lack robust forecasting models for future ice changes, which diminishes our ability to use these fingerprints to accurately predict local sea-level (LSL) changes. We exploit an advanced mathematical property of adjoint systems and determine the exact gradient of sea-level fingerprints with respect to local variations in the ice thickness of all of the world’s ice drainage systems. By exhaustively mapping these fingerprint gradients, we form a new diagnosis tool, henceforth referred to as gradient fingerprint mapping (GFM), that readily allows for improved assessments of future coastal inundation or emergence. We demonstrate that for Antarctica and Greenland, changes in the predictions of inundation at major port cities depend on the location of the drainage system. For example, in London, GFM shows LSL that is significantly affected by changes on the western part of the Greenland Ice Sheet (GrIS), whereas in New York, LSL change predictions are greatly sensitive to changes in the northeastern portions of the GrIS. We apply GFM to 293 major port cities to allow coastal planners to readily calculate LSL change as more reliable predictions of cryospheric mass changes become available.
How does network structure affect diffusion? Recent studies suggest that the answer depends on the type of contagion. Complex contagions, unlike infectious diseases (simple contagions), are affected by social reinforcement and homophily. Hence, the spread within highly clustered communities is enhanced, while diffusion across communities is hampered. A common hypothesis is that memes and behaviors are complex contagions. We show that, while most memes indeed spread like complex contagions, a few viral memes spread across many communities, like diseases. We demonstrate that the future popularity of a meme can be predicted by quantifying its early spreading pattern in terms of community concentration. The more communities a meme permeates, the more viral it is. We present a practical method to translate data about community structure into predictive knowledge about what information will spread widely. This connection contributes to our understanding in computational social science, social media analytics, and marketing applications.
Secondary use of electronic health records (EHRs) promises to advance clinical research and better inform clinical decision making. Challenges in summarizing and representing patient data prevent widespread practice of predictive modeling using EHRs. Here we present a novel unsupervised deep feature learning method to derive a general-purpose patient representation from EHR data that facilitates clinical predictive modeling. In particular, a three-layer stack of denoising autoencoders was used to capture hierarchical regularities and dependencies in the aggregated EHRs of about 700,000 patients from the Mount Sinai data warehouse. The result is a representation we name “deep patient”. We evaluated this representation as broadly predictive of health states by assessing the probability of patients to develop various diseases. We performed evaluation using 76,214 test patients comprising 78 diseases from diverse clinical domains and temporal windows. Our results significantly outperformed those achieved using representations based on raw EHR data and alternative feature learning strategies. Prediction performance for severe diabetes, schizophrenia, and various cancers were among the top performing. These findings indicate that deep learning applied to EHRs can derive patient representations that offer improved clinical predictions, and could provide a machine learning framework for augmenting clinical decision systems.
Correctly assessing a scientist’s past research impact and potential for future impact is key in recruitment decisions and other evaluation processes. While a candidate’s future impact is the main concern for these decisions, most measures only quantify the impact of previous work. Recently, it has been argued that linear regression models are capable of predicting a scientist’s future impact. By applying that future impact model to 762 careers drawn from three disciplines: physics, biology, and mathematics, we identify a number of subtle, but critical, flaws in current models. Specifically, cumulative non-decreasing measures like the h-index contain intrinsic autocorrelation, resulting in significant overestimation of their “predictive power”. Moreover, the predictive power of these models depend heavily upon scientists' career age, producing least accurate estimates for young researchers. Our results place in doubt the suitability of such models, and indicate further investigation is required before they can be used in recruiting decisions.
Thousands of lives are lost every year in developing countries for failing to detect epidemics early because of the lack of real-time disease surveillance data. We present results from a large-scale deployment of a telephone triage service as a basis for dengue forecasting in Pakistan. Our system uses statistical analysis of dengue-related phone calls to accurately forecast suspected dengue cases 2 to 3 weeks ahead of time at a subcity level (correlation of up to 0.93). Our system has been operational at scale in Pakistan for the past 3 years and has received more than 300,000 phone calls. The predictions from our system are widely disseminated to public health officials and form a critical part of active government strategies for dengue containment. Our work is the first to demonstrate, with significant empirical evidence, that an accurate, location-specific disease forecasting system can be built using analysis of call volume data from a public health hotline.
Will they participate? predicting patients' response to clinical trial invitations in a pediatric emergency department
- Journal of the American Medical Informatics Association : JAMIA
- Published over 2 years ago
(1) To develop an automated algorithm to predict a patient’s response (ie, if the patient agrees or declines) before he/she is approached for a clinical trial invitation; (2) to assess the algorithm performance and the predictors on real-world patient recruitment data for a diverse set of clinical trials in a pediatric emergency department; and (3) to identify directions for future studies in predicting patients' participation response.
Many spatial interpolation methods perform well for gentle terrains when producing spatially continuous surfaces based on ground point data. However, few interpolation methods perform satisfactorily for complex terrains. Our objective in the present study was to analyze the suitability of several popular interpolation methods for complex terrains and propose an optimal method. A data set of 153 soil water profiles (1 m) from the semiarid hilly gully Loess Plateau of China was used, generated under a wide range of land use types, vegetation types and topographic positions. Four spatial interpolation methods, including ordinary kriging, inverse distance weighting, linear regression and regression kriging were used for modeling, randomly partitioning the data set into 2/3 for model fit and 1/3 for independent testing. The performance of each method was assessed quantitatively in terms of mean-absolute-percentage-error, root-mean-square-error, and goodness-of-prediction statistic. The results showed that the prediction accuracy differed significantly between each method in complex terrain. The ordinary kriging and inverse distance weighted methods performed poorly due to the poor spatial autocorrelation of soil moisture at small catchment scale with complex terrain, where the environmental impact factors were discontinuous in space. The linear regression model was much more suitable to the complex terrain than the former two distance-based methods, but the predicted soil moisture changed too sharply near the boundary of the land use types and junction of the sunny (southern) and shady (northern) slopes, which was inconsistent with reality because soil moisture should change gradually in short distance due to its mobility in soil. The most optimal interpolation method in this study for the complex terrain was the hybrid regression kriging, which produced a detailed, reasonable prediction map with better accuracy and prediction effectiveness.
Using the DOSE index to predict changes in health status of patients with COPD: a prospective cohort study
- Primary care respiratory journal : journal of the General Practice Airways Group
- Published over 5 years ago
BACKGROUND: The severity of chronic obstructive pulmonary disease (COPD) should not be based on the level of airflow limitation alone. A multicomponent index such as the DOSE index (dyspnoea score (D), level of airflow obstruction (O), current smoking status (S), and exacerbations (E)) has the potential to predict important future outcomes in patients with COPD more effectively than the forced expiratory volume in one second. Health status deterioration should be prevented in COPD patients. AIMS: To investigate whether the DOSE index can predict which patients are at risk of a clinically relevant change in health status. METHODS: A prospective cohort study was performed using data from primary and secondary care. The DOSE score was determined at baseline and the 2-year change in the Clinical COPD Questionnaire (CCQ) score was calculated. Linear regression analysis was performed for the effect of a high DOSE score (≥4) on the change in CCQ score. RESULTS: The study population consisted of 209 patients (112 patients from primary care). Overall, a high DOSE score was a significant predictor of a change in CCQ score after 2 years (0.41, 95% CI 0.13 to 0.70), particularly in primary care patients. CONCLUSIONS: A DOSE score of ≥4 has the ability to identify COPD patients with a greater risk of future worsening in health status.
- Environmental health : a global access science source
- Published over 1 year ago
Despite the prediction that temperatures are expected to increase in the future, little is known about the health effects of increasing temperatures on pregnant women. The objective of this study was to investigate the impact of apparent temperature on spontaneous preterm delivery (PTD).
The sustainability of future water resources is of paramount importance and is affected by many factors, including population, wealth and climate. Inherent in current methods to estimate these factors in the future is the uncertainty of their prediction. In this study, we integrate a large ensemble of scenarios-internally consistent across economics, emissions, climate, and population-to develop a risk portfolio of water stress over a large portion of Asia that includes China, India, and Mainland Southeast Asia in a future with unconstrained emissions. We isolate the effects of socioeconomic growth from the effects of climate change in order to identify the primary drivers of stress on water resources. We find that water needs related to socioeconomic changes, which are currently small, are likely to increase considerably in the future, often overshadowing the effect of climate change on levels of water stress. As a result, there is a high risk of severe water stress in densely populated watersheds by 2050, compared to recent history. There is strong evidence to suggest that, in the absence of autonomous adaptation or societal response, a much larger portion of the region’s population will live in water-stressed regions in the near future. Tools and studies such as these can effectively investigate large-scale system sensitivities and can be useful in engaging and informing decision makers.