A Software Tool for Floating Point Interval Analysis with Improved Precision for Javascript-Based Medical Applications

Medical applications intimately rely on floating point numbers to store, retrieve and process important physical as well as diagnostic data associated with medical clients. The occurrence of unexpected loss of precision over floating point numbers under medical contexts can cause great damage, or even cost lives. Due to its excellent interoperability with web-based content, Javascript has been an important source language to craft mission-critical medical applications utilizing the web-based capabilities. For those applications, applying javascript static program analysis is an important counter measure against runtime errors including floating point number errors. In this paper, we propose an abstract interpretation based interval analysis for Javascript programs using floating point bounds. We base our method and results on top JSAI, a Javascript static analysis framework.