The most important aspect of SleepXML is defining the sleep events with the XML Schema. XML Schema is a companion standard that provides a way to better describe the structure and permissible content of an XML do*****ent (http://www.w3c.org/XML/Schema). SleepXML Schema uses standard Schema Definition Language (XSDL) to: –define PSG channel types (Airflow, SpO2, etc.); –define PSG events on those channels (respiratory, arousals, etc.); –define event parameters (time, duration, SpO2 drop, stage, position). Event characteristics, such as SaO2 drop for a hypopnea, can be defined for specific events. Characteristics of event display, such as screen color, and scoring preferences (e.g. shortcut keystrokes) are also defined via the SleepXML attributes. Figure 1 presents fragment of the SleepXML Schema (rendered in Internet Explorer) that defines Respiratory Events according to the structure used in NYU SDC. This lab has a number of classifications for the Respiratory event, such as Central Apnea, Obstructive Apnea or Hypopnea, enumerated here as XML elements.
Figure 1. fragment of the SleepXML Schema defining Respiratory Events.

The PSG events data in the SleepXML format corresponded to this schema is depicted on the Figure 2:

Figure 2. The PSG events correspondent to the Figure 1 Schema, rendered in Internet Explorer. Other Sleep labs may decide to store other elements or parameters of their PSG Events according to their implementation of SleepXML schema. If we have two different Schemas defining the PSG events, we can transform one set of events to another by utilizing another XML standard – XSLT. XSLT provides a mechanism to allow easy manipulation of XML do*****ents. There are a number of commercial and Open Source XSLT processors available (e.g. Apache’s open source Xalan, open-source Saxon, Microsoft XML tools, Oracle’s XML parser). To test the concept, we have prepared XSLT templates and executed the transformation of SleepXML events from those defined according to Chicago Criteria [3] to another set of events conformant to Medicare Criteria [4]– see Figure 3. Another transformation that we executed was from the NYU SDC format to the PhysioBank-based format [5]. Such transformations could be done automatically based on two Schema-based event definitions.

Figure 3. Fragment of XSLT template for conversion of Chicago Criteria events to Medicare events.