>8 ; OVER 14 CLAIMS MORE LIKELY INVOLVED IN PATENT LITIGATION !
in response to
by
posted on
Oct 25, 2008 10:24PM
David E. Martin, PhD∞
Peter A. Beling, PhD†
Using the CART model, classification performance improved to correct prediction of litigation or non-litigation control 72.4% of the time. Litigated patents were correctly classified 55.8% of the time and non-litigated patents were correctly classified 89% of the time. As with the logistic regression model, the CART model showed that the single most influential predictor of involvement in litigation was the existence of N1 ratings.
The relative risk of being involved in litigation for several significant risk factors is summarized in Table 1. By relative risk we mean the ratio of the probability of being involved in litigation to the probability of not being involved in litigation. This ratio is approximately 1.1 for the complete set of patents used in the study, since roughly equal numbers of litigated non-litigated patents were chosen for inclusion. From the table it can be seen that a patent with an N1 icon is more than ten times as likely to be involved in litigation as it is to be not involved in litigation. Similarly, a patent with fewer than 8 claims is about half as likely to be involved in litigation as it is to be not involved in litigation.
Table 1. The unique contribution of isolated patent risk variables as they impact the relative risk of being involved in patent litigation
Risk Factor |
Relative Risk of Being Involved in Litigation |
N1 rating |
10.3 |
Number of claims >26 |
2.1 |
Number of claims < 8 |
0.54 |
Number of claims <26 but >8; over 14 subsequent art and, over 30 prior art references |
9.0 |
Discussion
The ability to detect patents that are at risk for patent litigation can be accurately assessed using the variables measured in the present study. Adverse selection risk can be significantly ameliorated using the model presented here. Insurance and impairment assessment that fails to consider patent uniqueness will lead to avoidable losses.
These results unequivocally demonstrate that possession of certain patent properties can increase the risk for a company or individual being involved in patent litigation.