« 上一頁繼續 »
TRS to buyer ond seller by phase
Percentage of Excess TRS 3 T 2.5 n 2.1 2 _ 1.5 — 1.2 1.3 \ 1.3 \ 1 _ 0.6 0.5 — / no 02 \ 0 -0.7 1/ ‘ /I --0.5 — _1 _ n=66 n:39 n=95 n=70 n:67 n=64 n=228 n=173 Phase I Phase II Phase III Grond Totol Excess TRS Buyer , Figure 4
One of the opproved for Bottom 25% TRE Buyer wos excluded from the lineor regression due to 0 corporate oction in the 11-day period.
1 SYSTEM FOR PREDICTIVE ANALYTICS USING REAL-WORLD PHARMACEUTICAL TRANSACTIONS
CROSS-REFERENCE TO RELATED APPLICATIONS
This application claims the benefit of priority from copending provisional patent application Ser. No. 61/049,914, filed on May 2, 2008, entitled System For Predictive Analytics Using Real-World Pharmaceutical Transactions. Application Ser. No. 61/049,914 is hereby incorporated by reference in its entirety.
BACKGROUND OF THE INVENTION
1. Technical Field
This disclosure relates to predictive analytics in the pharmaceutical industry. In particular, this disclosure relates to predicting the likelihood of success of a real-world pharmaceutical transaction based on analysis of prior pharmaceutical transactions.
Pharmaceutical companies often venture outside of their own organizations in search of new assets to develop in their pipeline and to ultimately bring to market. These deals may be in the form of a purchase, license, joint development, strategic arrangement, or other business transaction. However, the key factors and conditions that predict and quantify a successful business transaction or relationship have not been established. Typically, pharmaceutical industry experts have relied on relationships, qualitative evidence, intuition, or experience-based “rules of thum ” when establishing licensing programs or collaborative arrangements. However, there has been no quantitative evidence to prove that any of these techniques are successful in selecting deals that create more value. Reliance on such factors does not necessarily increase the probability that the transaction will be successful. An unmet need exists to identify and quantify the factors and conditions that correspond to successful pharmaceutical business transactions or relationships.
The pressure on pharmaceutical companies to achieve high performance and deliver new products has never been greater. With revenues eroding because of expiring patents and generic competition, companies are in a collective scramble to acquire new compounds. In search of the next innovation, companies have been through a decade of whirlwind dealmaking to bring in products from external sources. Whether through licensing or more elaborate business development investments, the number and value of these deals are only expected to increase.
However, only a small portion of these deals result in successful products. To increase the success of these efforts to feed the pipeline, companies have two strategic options. The pharmaceutical companies can either engage in more deals, or they can become increasingly selective and engage in fewer deals. The pharmaceutical industry now relies heavily on the first strategy-raising its level of investment in the hope of yielding a higher absolute number of successful products. Because there are limited resources for investing in new deals, there is a need for a tool that can assist pharmaceutical companies to be more selective, and to engage in fewer deals that have a higher probability of successful returns.
The system and method for predictive analytics using realworld pharmaceutical transactions addresses a second option,
namely increasing selectivity in the business development deals a company makes, by identifying the characteristics that make a deal most likely to succeed. One embodiment of a system for predictive analytics using real-world pharmaceutical transactions includes a computer having a processor and memory, a data collection component configured to aggregate data for a plurality of pharmaceutical transactions where the aggregate data corresponds to publicly-traded financial data based upon a predetermined time period surrounding a public announcement of the respective pharmaceutical transaction.
A data reduction module eliminates non-compliant transactions to generate a reduced transaction data set and an analysis module applies multiple linear regression analysis to a portion of the reduced transaction data set to identify key regression variables that correlate with an excess total shareholder return. The key regression variables that were statistically significant in this analysis were drug development phase, deal type, compound or drug type, and therapeutic area. The analysis module also applies logistic regression analysis to a portion of the reduced transaction data set to identify key regression variables that correlate with an increased probability of regulatory agency approval. A report generator provides a graphical output of the identified key regression variables and a probability value corresponding to a likelihood of regulatory agency approval.
Other embodiments of systems, methods, features, and their corresponding advantages will be, or will become, apparent to one with skill in the art upon examination of the following figures and detailed description. It is intended that all such additional systems, methods, features, and advantages be included within this description, be within the scope of the invention, and be protected by the following claims.
BRIEF DESCRIPTION OF THE DRAWINGS
The system may be better understood with reference to the following drawings and the description, in addition to the presentation sheets included in the appendix, which is incorporated herein in its entirety. The components in the figures are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the invention. Moreover, in the figures, like-referenced numerals designate corresponding parts throughout the different views.
FIG. 1 shows a computing platform and environment;
FIG. 2 is a flowchart showing a process identifying key regression variables that correlate with an excess total return to shareholders;
FIG. 3 is a flowchart showing a process predicting the likelihood of eventual FDA approval;
FIG. 4 is a bar chart showing median excess TRS to buyer and seller by clinical phase;
FIG. 5 is a graph illustrating the predictive quality of the stock market for phase III compounds; and
FIG. 6 is a graph illustrating the predictive quality of the stock market for phase II compounds.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
As shown in FIG. 1, a system 100 for predictive analytics using real-world pharmaceutical transactions provides a platform for applying a statistically rigorous process that identifies “deal” or transaction characteristics that are likely to predict success. The specific embodiment of FIG. 1 is a highlevel hardware block diagram of a computer system on which the system 100 for predictive analytics using real-world pharmaceutical transactions may be implemented. The system
100 for predictive analytics using real-world pharmaceutical transactions may be embodied as a system cooperating With computer hardware components and/or as a computer-implemented method.
The system 100 includes a predictive analytic engine or processor 102, Which in turn, includes an analysis module 104 or processor, a comparator module 105, a data collection component 106, a data reduction module 108, and a report generator 109. The predictive analytic engine 102 may be a hardware component and/or may performed processes in hardware, software, or a combination of hardware and software. The system 100 includes a computer or processing system 112, which includes various hardware components, such as RAM 114, ROM 116, hard disk storage 118, cache memory 120, database storage 122, and the like (also referred to as “memory subsystem” 126). The computer system 112 may include any suitable processing device 128, such as a computer, microprocessor, RISC processor (reduced instruction set computer), CISC processor (complex instruction set computer), mainframe computer, work station, single-chip computer, distributed processor, server, controller, microcontroller, discrete logic computer, and the like, as is known in the art. For example, the processing device 128 may be an Intel Pentium® microprocessor, x86 compatible microprocessor, or equivalent device.
The memory subsystem 126 may include any suitable storage components, such as RAM, EPROM (electrically programmable ROM), flash memory, dynamic memory, static memory, FIFO (first-in first-out) memory, LIFO (last-in firstout) memory, circular memory, semiconductor memory, bubble memory, buffer memory, disk memory, optical memory, cache memory, and the like. Any suitable form of memory may be used whether fixed storage on a magnetic medium, storage in a semiconductor device, or remote storage accessible through a communication link. A user or system manager interface 130 may be coupled to the computer system 112 and may include various input devices 136, such as switches selectable by the system manager and/or a keyboard. The user interface also may include suitable output devices 140, such as an LCD display, a CRT, various LED indicators, and/ or a speech output device, as is known in the art.
To facilitate communication between the computer system 112 and external sources, a communication interface 142 may be operatively coupled to the computer system. The communication interface 142 may be, for example, a local area network, such as an Ethernet network, intranet, Internet, or other suitable network 144. The communication interface 142 may also be connected to a public switched telephone network (PSTN) 146 or POTS (plain old telephone system), which may facilitate communication via the Internet 144. Dedicated and remote networks may also be employed, and the system may further communicate with external exchanges and sources of information 146.Any suitable commercially-available communication device or network may be used.
FIG. 2 is a flowchart showing a process (Act 200) that identifies variables that that correlate with an excess total return to shareholders. The process shown, for example in FIG. 2, may be performed by the predictive analytics engine 102. Data for a plurality of pharmaceutical transactions is collected (Act 210), and non-compliant transactions are eliminated (Act 220) to generate a reduced transaction data set. Multiple linear regression analysis is applied to a portion of the reduced transaction data set (Act 230) to identify key regression variables that correlate with an excess total return to shareholders (Act 240). A report is output (Act 250) that
provides a graphical output of the identified key regression variables and an indication of the corresponding effect on total return to shareholders.
FIG. 3 is a flowchart showing a process (Act 300) that determines a probability that a pharmaceutical deal will result in eventual FDA approval. The process shown, for example in FIG. 3, may be performed by the predictive analytics engine 102. Data for a plurality of pharmaceutical transactions is collected (Act 310), and non-compliant transactions are eliminated (Act 320) to generate a reduced transaction data set. Logistic regression analysis is applied to a portion of the reduced transaction data set (Act 330) to identify key regression variables that correlate with an increased probability of obtaining eventual FDA approval (Act 340). A report is output (Act 350) that provides a graphical output of the identified key regression variables and a probability value corresponding to a likelihood of regulatory agency approval.
Several factors or predictors can indicate which external sourcing arrangements or deal has the best prospects of becoming a winner. In one specific embodiment, the term “success” may be identified based on two separate criteria. A first measure of success is defined as a deal that increases shareholder value (excess total return to shareholders (TRS) or ETRS). “Excess” total return to shareholders may be defined as the 1 1-day compounded TRS (as defined by CRSP) minus the 11-day compounded TRS of a corresponding index. In one embodiment, the normal return is defined as the TRS AMEX pharrna for a pharmaceutical company and AMEX biotech for a biotech company. An 1 1-day window is a common time frame used in financial transactions because 11 days is believed to be an adequately long window for capturing the market’s reaction to the event, namely the announcement of the deal, while still being sufliciently short so as to limit the impact of other corporate events. Data points may be omitted where other major events occurred during the 11-day window. Other timeframes may be used, such as a week, a month, a quarter, or a year.
A second measure of success is defined as a deal that results in eventual FDA (Food and Drug Administration) or other regulatory body approval of the drug compound that is the subject of the deal. Preferably, deals where the drug under evaluation are approved or terminated are analyzed. Deals with ongoing research or pending FDA approval are excluded. Analysis (logistic regression) for Phase I deals are excluded because few Phase I compounds have received approval from the FDA.
In one embodiment, the predictive analytics engine 102 applies regression analysis (multiple linear regression and logistic regression) to identify factors that correlate with a deal’ s success. Many factors were investigated and analyzed. Such factors may include:
1. Company Acquiring Asset Party Type (CAA)
2. Company with Asset Party Type (CWA)
3. Parties CAA/CWA
4. New Deal Type
6. Weakness of Incentive (UpfrontAmount divided by Deal
7. Stage of Compound(s)
8. Therapeutic Area
9. Biologic/Small Molecule/Other
10. Transaction # with same partner
1 1. Prior Relationship with Partner
12. Total Transaction Frequency CAA
13. Total Transaction Frequency CWA
The predictive analytics engine 102 identified four predictors of a deal’s success based on regression analysis of the