Today's Internet makes hosts and individual networks inherently insecure because permanent addresses turn destinations into permanent attack targets. This paper describes an Evasive Internet Protocol (EIP), a change to the data plane of the Internet that: 1) prevents senders from forging their identities while preserving the current Internet privacy paradigm; 2) gives recipients full control over who can communicate with them and for how long; 3) achieves the above features without requiring global signaling protocols; and 4) allows coexistence with and graceful introduction into the current Internet. We motivate our approach, present the architectural design, and evaluate it through trace-driven and synthetic simulations as well as prototype testing.

Diagnostic medicine produces large volumes of free-text reports used primarily for communication between medical professionals. Secondary use of these reports requires extraction of structured information from the free text. State-of-the-art computational natural language processing techniques can make partial identification of semantics in text, but the diverse terminology used in medical settings makes training classifiers for every lexicon a laborious task. We present statistics of semantics from a large-scale machine-annotated corpus of 83,452 chest x-ray reports. We show that the distribution of semantics is consistent with Zipfian distributions observed in other natural language corpora, and we quantify the semantic focus imparted by limiting a study by body area and modality. We demonstrate that within our semantically focused corpus, pairwise co-occurrence statistics can be used to accurately impute the semantic class for frequently occurring unknown entities, thereby reducing the number of semantically unclassified phrases by up to 25%. Finally, we show that our imputation approach is consistent across multiple reconstructions of the underlying text data.

Diagnostic radiology reports contain a summary written by a radiologist for communication with primary care providers. Information not included in the summary may be lost in the communication, resulting in substandard patient care. We introduce a notion of salience based on distributional characteristics of term usage in a corpus of radiology reports and present an algorithm for generation of suggestions for inclusion of additional information in report summaries. We evaluate our method on a corpus of 98,913 reports and show our method suggests additions to 11-28% of reports, depending on body location and imaging modality.