Refinement and enhancement of a web-based risk calculator deployment system .The quality of medical decision making suffers because risk predictions are not tailored to the individual patient. Further, available risk models and tools are not widely used because they are not easy to use, not useful, or both. Having the ability to provide bedside predictions from statistical models that permit continuous risk estimation would represent major progress and has the potential for broad and significant impact on how clinicians and patients make treatment decisions. We have begun to develop a web-based infrastructure that allows statisticians to quickly produce an online risk calculator to facilitate application of prediction models. The goal of this proposal is to enhance and broadly expand our risk calculator constructor, which is designed to provide risk estimations tailored to the individual patient. No other generic online risk calculator is available. The infrastructure is designed to rapidly accelerate the translation of risk equations to clinical practice by making the process very easy for a nonprogrammer to create a user- friendly risk calculator application. The tool is free to both statistician and end-user (e.g., clinician), easy to use by both parties, and runs on any web browser (including the BlackBerry's). Although we have a working beta version, we need to enhance and expand it considerably before it will gain widespread acceptance. Therefore, we aim to 1) extend the existing functionality. Both the authoring and clinician interfaces would benefit from additional functionality, including support for a wider variety of input variable formats and more expressive validity constraints, inclusion of additional predefined statistical models, extended documentation and annotation capabilities, provision of an """"""""undo"""""""" feature, and more flexible publishing capabilities. Most of these enhancements involve changes to the existing editing tool and their corresponding calculator generation and evaluation components. 2) develop new functionality. We will change the underlying system architecture so it will be capable of on-the-fly imputation of missing values present when performing risk calculations, as well as arranging multiple predictions in a matrix of results to generate comparative effectiveness tables, among other functions. 3) revise the underlying infrastructure and improve robustness. Minor tweaks are needed to smooth system performance, including testing some newly introduced features and revisions using a more systematic methodology to assure consistency, adding support for a wider variety of platforms, and more intelligent use of available web-browser capabilities. We would like to develop the system's surveillance capabilities to capture and analyze pertinent statistics on calculator use and tend to observed problems. In summary, this technology is highly innovative and can be applied to many medical decisions, which is a very broad impact. It has the potential to radically affect almost any area of medical decision making where statistical models can be developed.
When faced with complex medical decisions, physicians and patients could make much better decisions if predicted outcomes were tailored to the individual patient. No cost-effective method is available that permits this precise tailoring. We are developing an easy-to-use online tool that allows patient outcome to be predicted based on the patient's individual characteristics. If our tool is widely adopted, it has the potential to radically affect medical decision making and to have a broad impact on how all physicians practice medicine.
