Claim Analytics v. Polon et al., 2016 ONSC 2235

CITATION: Claim Analytics v. Polon et al., 2016 ONSC 2235

COURT FILE NO.: CV-16-544562

DATE: 20160408

ONTARIO

SUPERIOR COURT OF JUSTICE

BETWEEN:

CLAIM ANALYTICS INC.

Plaintiff

– and –

JONATHAN POLON and ACTUARIAL INTELLIGENCE INC.

Defendants

Matthew Morden, for the Plaintiff

Norman Ronski, for the Defendants

HEARD: March 22, 2016

LEDERMAN J.

NATURE OF MOTION

[1] The plaintiff seeks an interim and interlocutory injunction against the defendants to prevent the disclosure of confidential information, being the “Diagnosis Database”. It is a predictive model that the plaintiff uses in its core business to serve a very narrow area of demand within a specific insurance market, namely group disability.

BACKGROUND

[2] The plaintiff is a corporation founded by the defendant, Jonathan Polon (“Polon”) and Barry Senensky (“Senensky”).

[3] The plaintiff uses data, computing power, and the expertise it has developed since it was incorporated in 2002 to create predictive models which provide claim scoring services, in relation to, among other things, short-term disability and long-term disability insurance claims. The scores indicate the likelihood that a claimant will return to work within a period of time. Insurers can use these models to make decisions about how to handle claims and to measure their own performance as against their competitors.

[4] Senensky’s prior experience had been in the insurance industry whereas Polon’s was in the creation of predictive modelling. They concluded that there might be a demand in the insurance industry for predictive analytics and models. Polon applied his expertise and skill in the areas of statistical analysis and predictive modelling to develop and oversee the development of the plaintiff’s technical approaches, processes and models.

[5] The process by which Claim Analytics built these first predictive models was:

(a) obtain a dataset of historical disability claims;

(b) prepare the data for modelling. This includes not only checking for and removing bad data, but also, most importantly, transforming raw data into forms that can be used by a predictive model. Among other things, Claim Analytics devised an innovative way to transform medical diagnoses into numerical scores (the “Diagnosis Database”); and

(c) Follow a method to build and optimize statistical needs. The transformed data was used to build these statistical models.

[6] In March, 2014, after Senensky triggered a buy/sell provision in the Shareholders Agreement entered into by their respective shareholder corporations, Polon sold his interest in the plaintiff to Senensky for $2,500,000 and left Claim Analytics.

THE DIAGNOSIS DATABASE

[7] The Diagnosis Database is a key part of the plaintiff’s confidential information. Its value lies in its use in transforming a particular medical diagnosis code to numerical variables which assist in predicting the likelihood of a claimant’s return to work. The Diagnosis Database tool allows potentially unrelated diagnoses to be grouped together which facilitates more accurate statistical modelling.

BENCHMARKING STUDIES

[8] In addition to creating and offering predictive models in respect of long-term and short-term disability claim scoring, another core part of the plaintiff’s business is to provide services in relation to “Benchmarking Studies”, which provide a claim recovery comparison to various insurers. The objective of a Benchmarking Study is to make use of the plaintiff’s statistical and predictive modelling tools to allow insurance company participants to see how their claim recovery experience compares to their competitors.

[9] The plaintiff and Munich Re Insurance Company Canada Branch (“Munich Re”) co-sponsored two Benchmarking Studies in relation to group long-term disability claims experience for Canadian insurers in 2011 and 2014.

[10] The plaintiff and Munich Re had discussed co-sponsorship of a new study to take place in 2016 and had planned to offer it to the Canadian insurers who participated in the previous study and to others.

[11] However, when the plaintiff insisted upon looking for some kind of guaranteed revenue from Munich Re, Munich Re advised the plaintiff that it was not acceptable and minimum revenues did not fit with Munich Re’s objectives. Munich Re then sought out Polon and agreed to conduct its 2016 Benchmarking Study with him rather than with the plaintiff, mainly because Polon was willing to do the study for the same financial structure as had been charged in the past.

[12] The plaintiff now seeks to enjoin Polon from providing such services to Munich Re or any other participant in relation to the proposed 2016 Canadian Benchmarking Study.

TEST FOR INJUNCTIVE RELIEF

[13] The plaintiff must satisfy each of the three branches of the test set out in RJR MacDonald Inc. v. Canada 1994 117 (SCC), [1994]1 S.C.R. 311, namely: a) that there is a serious question to be tried; b) that the applicant would suffer irreparable harm if the application were refused; and c) that the balance of convenience favours the applicant in that it would suffer greater harm from the refusal of an injunction pending a decision on the merits than would the respondents if an injunction were granted.

SERIOUS ISSUE TO BE TRIED

[14] Polon contends that the plaintiff’s real purpose is not to protect trade secrets and confidential information, but to prevent him from competing in any way. He submits that the plaintiff is seeking to effectively put Polon out of business which he has been carrying on unimpeded and without any attempt at restriction by the plaintiff since he sold out his interest in the plaintiff in March, 2014.

[15] The dispute centres around the confidentiality of the concept and technique to create a Diagnosis Database. Polon submits that the predictive models were unique 10 years ago but that is no longer the case. He states that all material confidential information and trade secrets have already been disclosed into the public domain by the plaintiff itself. This includes both a) information disseminated by the plaintiff in papers, articles and presentations as well as b) concepts developed by persons with no affiliation to the plaintiff inclusive of works published in text books, scholarly articles and conveyed to the public during presentations. Accordingly, qualified individuals are capable of building their own predictive models, including insurance claim scoring.

[16] With respect to the Diagnosis Database, it should be noted:

(a) the Confidential Information relates to specific novel methods, processes and compilations of information created and developed by the plaintiff;

(b) the plaintiff has consistently treated this information as confidential:

(i) although the general concept has been disclosed outside of the business, important elements of the Diagnosis Database remain secret;

(ii) the plaintiff is a closely-knit business with few employees;

(iii) the plaintiff has required third parties, employees, and even its shareholders to agree to the confidentiality of its information;

(iv) the Diagnosis Database represents an important data transformation which distinguishes the plaintiff’s predictive models. Polon knows this and has recorded in emails his opinion as to the value of this transformative process;

(v) the Diagnosis Database has been evolving and growing since 2003. In order to build a database which is relevant, the plaintiff has benefitted from the data provided to it by its customers;

(vi) Without specific knowledge of the seven variables for each of the diagnosis codes, the manner in which scores were assigned to the variables, and particularly the eighth element of the Diagnosis Database (which represents an ingenious twist on the data transformation), it cannot be easily duplicated.

[17] The plaintiff has mapped each medical diagnosis it has encountered in its customers’ data into a set of variables which are rated 0 or 1 to 10. The data sets were received from clients between 2002 and 2016. As of March 2014, there were seven variables scored for each of the hundreds of diagnosis codes. Of the seven variables, five have been disclosed in presentations or academic papers. Two of the variables remain confidential.

[18] Polon himself, in his Affidavit, admits that all the elements of the Diagnosis Database have not been put into the public domain. At para. 97 of his Affidavit he states as follows:

These papers and presentations do not disclose all of the values which are included in Claim Analytics diagnosis database but they do divulge 5 of the 8 values (plus some others not used by Claim Analytics), and, the remaining 3 values are ones that:

(1) I expect a person with expertise in disability claims management or rehabilitation could reasonably be expected to consider; and

(2) are likely not that material to the accuracy of a model.

The concept of encoding diagnosis from a categorical variable to a set of metrics is important. But I do not believe that the specific set of metrics chosen, so long as it is guided by a person with expertise in disability claims management, would be of material importance.

[19] Polon further agrees that there is an eighth element to the Diagnosis Database that has not been disclosed. That element apparently is a concept known as the theory of credibility which is an important part of actuarial science. Polon held the view at one time that this element “holds tremendous value” and allowed the plaintiff to succeed in “training a model that is far superior to [the insurers] underwriter ratings” ( see Exhibit L to Senensky’s Affidavit). Polon also stated in an email to Senensky on June 1, 2013, (Exhibit J) that this eighth element is “unlike the other elements because it is a calculated value rather than a subjective value. I think it is probably patentable. It is also, by far, the most important element in the table.”

[20] As stated by Ronald E. Dimock in Intellectual Property Disputes, Resolutions and Remedies, Volume 2, Toronto, Thomson Rueters 2004, at page 5-41

“where the confidential information consists of information available to the public, courts have held that there must be some product of the human brain that suffices to confer a confidential nature upon the information, whether it be described as originality, or novelty or ingenuity or otherwise”.

Thus, where public information is applied in a new way, it can be considered confidential information.

[21] Even when all of the information becomes public, if an ex-employee is able, by information provided by or developed for the previous employer to gain an advantage that he or she would not have had, it would provide a spring board for an unfair advantage over others. What is really being protected in situations of this nature is the original process of mind. The protection is enforced against persons who wish to use the confidential information without spending time, trouble and expense of going through the same process to prevent them from getting an unfair start: see Catalyst Capital Group Inc. v. Moyse 2014 ONSC 6442 at para. 18.

[22] In any event here, the full Diagnosis Database has not been disclosed publicly and the plaintiff has taken protective measures to ensure its confidentiality.

[23] With respect to the serious question to be tried, the threshold is low; the application must not be frivolous or vexatious.

[24] A misappropriation of trade secrets by a former employee will give rise to liability for which there is injunctive relief.

[25] There is sufficient evidence to establish that there is a serious issue to be tried with respect to the imminent threat of breach of confidence and fiduciary duty by Polon.

IRREPARABLE HARM

[26] Polon contends that the plaintiff has failed to provide any evidence whatsoever how it will suffer harm economically or otherwise if a 2016 Benchmarking Study is performed without its involvement, as Munich Re has clearly taken the position that it would not co-sponsor the study with the plaintiff because of the financial guarantee demanded by the plaintiff.

[27] However, what is at stake is not just the loss of a co-sponsorship with Munich Re. The plaintiff may wish to seek another co-sponsor or do the 2016 Benchmarking Study on its own.

[28] Unfair competition can lead to irreparable harm that cannot be compensated for in damages as it is difficult to quantify the loss of goodwill or market share suffered by a plaintiff due to a defendant’s actions, particularly in a competitive industry where there can be considerable fluidity of customer allegiances; see Catalyst supra, at para. 76.

[29] What is clear is that Munich Re seeks a co-sponsorship with Polon rather than doing the 2016 Benchmarking Study on its own. The plaintiff’s legitimate fear is that Polon will use and disclose the Diagnosis Database to Munich Re or to other employees or third parties.

[30] In this way, the plaintiff has established that it may well suffer irreparable harm if an injunction were not granted.

BALANCE OF CONVENIENCE

[31] If no injunction is granted, the defendants would be at liberty to use the plaintiff’s Diagnosis Database and in the interim prior to trial, the plaintiff may lose the opportunity to conduct its own 2016 Canadian Benchmarking Study due to the defendants’ unfair competition.

[32] The limitation on Polon, should an injunction be granted, is that he will be precluded from using the plaintiff’s Diagnosis Database in any co-sponsorship with Munich Re or with others. He, however, has actuarial skills which can be applied beyond the group insurance market and there are many other sectors of the insurance industry to which Polon may provide his expertise. It should be remembered that it was Polon who once indicated that the plaintiff’s most valuable asset (the “Diagnosis Database”) was its intellectual capital and sold his interest to his former partner and co-founder for $2,500,000. It would not be inequitable to insist that he not use that “intellectual capital” to duplicate the plaintiff’s business in the interim period prior to a trial.

[33] The plaintiff will suffer greater harm if Polon discloses its trade secrets and confidential information than Polon would if he were restrained from using them. The balance of convenience therefore favours the granting of injunctive relief.

CONCLUSION

[34] Accordingly, an order will go restraining the defendants until the trial of the action from using in any way, directly or indirectly, on their own behalf or on behalf of any other person, the plaintiff’s Diagnosis Database comprised of the seven metrics developed by Polon and utilizing the eighth common diagnosis - related element which is a calculated value arrived by utilizing the theory of credibility as described in paragraphs 26 to 28 in Polon’s Affidavit and that would include restraining the defendants from using the plaintiff’s Diagnosis Database in any services provided to Munich Re or any other participant in relation to a proposed 2016 Canadian Benchmarking Study.

[35] Given the necessity to protect the confidentiality of commercially sensitive material, a sealing order will go with respect to the Motion Record – Part II, the Responding Record of the Defendant and the Supplementary Factum of the Defendants.

[36] If the parties cannot otherwise agree as to costs of the motion, they may make written submissions within 30 days.

Lederman J.

Released: April 8, 2016