A pharmaceutical enterprise innovation service system

Abstract

The application relates to the technical field of medical service systems, and discloses a medical enterprise innovation service system which comprises a data acquisition module, a data analysis module, an innovation potential evaluation module, an intelligent recommendation module, a result feedback module, a dynamic weight optimization module, a threshold self-adaption module, a multi-target recommendation module, a cooperation value evaluation module, a deep learning optimization module and a block chain cooperation module. Through technology fusion and process innovation, the application solves the problems of traditional systems in data processing, dynamic adaptation, multi-target recommendation and data security and the like, and provides an intelligent, automatic, safe and reliable innovation service solution for medical enterprises.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical service systems, in particular to a medical enterprise innovation service system. BACKGROUND

[0002] In the process of innovation and development of medical enterprises, accurate innovation capability evaluation and efficient strategy recommendation are the key to improving core competitiveness. Traditional enterprise innovation service systems mainly rely on manual experience or simple data analysis models, which have the following technical bottlenecks:

[0003] 1. Insufficient data processing and analysis capabilities: The existing system has limited integration and depth analysis capabilities for multi-source heterogeneous data (such as R&D investment, market data, patent information, etc.), lacks feature extraction and time series analysis based on deep learning, and is difficult to mine potential correlations between data, resulting in inaccurate innovation potential evaluation.

[0004] 2. Lack of dynamic adaptability: Weight setting and threshold judgment are mostly static configurations, which cannot be dynamically adjusted according to actual feedback from enterprises and market changes. For example, the weights of innovation investment, market performance, and patent competition cannot be optimized in real time, and the threshold setting lacks self-adaptability, resulting in low matching degree between recommended solutions and actual needs of enterprises.

[0005] 3. Limitations of multi-objective recommendation and cooperation evaluation: Traditional systems have deficiencies in comprehensive recommendation of multi-dimensional needs (such as funds, time, patent protection, and cooperation strategies), and lack historical data support and automated processes for evaluating the value of potential partners, making it difficult to achieve efficient matching and execution of cooperation processes.

[0006] 4. Lack of data security and process automation: Data storage relies on centralized storage, which has the risk of tampering; the execution of cooperation solutions lacks automated mechanisms, and manual intervention costs are high, making it difficult to ensure the transparency and efficiency of the process. SUMMARY

[0007] To overcome the deficiencies of the prior art, the present application provides a medical enterprise innovation service system, which solves the problems in the background art.

[0008] To achieve the above purpose, the present application is realized by the following technical scheme: a medical enterprise innovation service system, characterized by comprising a data acquisition module, a data analysis module, an innovation potential evaluation module, an intelligent recommendation module, a result feedback module, a dynamic weight optimization module, a threshold self-adaptation module, a multi-objective recommendation module, a cooperation value evaluation module, and a deep learning optimization module and a blockchain collaboration module:

[0009] The data acquisition module is used to acquire multi-source data from internal and external channels of the enterprise and store them as a set , wherein , represents enterprise internal data, represents external data; the deep learning optimization module is connected with the data collection module, the data analysis module, the innovation potential evaluation module and the intelligent recommendation module, uses a convolutional neural network and a long short-term memory network to perform feature extraction and time series analysis on the collected multi-source data, and mines potential correlations between the data; the data analysis module is used for cleaning and preprocessing the collected data to obtain clean data , and extracting key indicators from to integrate into a vector and transmit to the innovation potential evaluation module; the innovation potential evaluation module is used for combining innovation input factors , market performance factors , patent competition factors and corresponding weights , , , calculating a comprehensive innovation potential score , and transmitting to the intelligent recommendation module and the threshold self-adaptive module; the intelligent recommendation module is used for generating a recommended scheme set based on and a knowledge base; the result feedback module is used for collecting feedback information after the enterprise executes the recommended scheme, obtaining a feedback factor after normalization processing, and transmitting to the dynamic weight optimization module, the threshold self-adaptive module, the multi-objective recommendation module and the cooperation value evaluation module; the dynamic weight optimization module is used for dynamically updating the weights in the innovation potential evaluation module based on ; the threshold self-adaptive module is used for dynamically updating the threshold value according to the difference between and the current threshold value ; the multi-objective recommendation module is used for calculating a multi-objective recommendation index and generating a comprehensive recommended scheme based on in combination with the needs of the enterprise in the dimensions of funds, time, patent protection and cooperation strategy; the cooperation value evaluation module is used for calculating a cooperation value index for potential cooperation parties, and outputting an optimal cooperation combination in combination with when a partner is selected or joint research and development is involved; the blockchain collaboration module is connected with the data collection module, the result feedback module and the cooperation value evaluation module.

[0010] Preferably, the data analysis module comprises a data cleaning and preprocessing unit and a key indicator extraction unit;

[0011] The data cleaning and preprocessing unit uses a transformation function. The multi-source data acquired from the data acquisition module is processed by deduplication, missing value imputation, outlier removal, or normalization to obtain... ,in, Transformation function for the original dataset Adjusted by the optimization parameters output by the deep learning optimization module;

[0012] The key indicator extraction unit Extracting R&D investment intensity Market share Number of patents Patent influence Key metrics, integration vector This vector will serve as input for the subsequent innovation potential evaluation module.

[0013] Preferably, in the innovation potential evaluation module, the system combines innovation input factors. Market performance factors Patent competition factors and corresponding weights , , Calculate the comprehensive innovation potential score ;

[0014] The specific calculation method is as follows:

[0015] Innovation input factors: ,in It is a function defined according to enterprise needs and industry characteristics, and is adjusted by the optimization strategy output by the deep learning optimization module;

[0016] Market performance factors: ,in Similarly, it is a function defined according to enterprise needs and industry characteristics, and is affected by the optimization strategy of the deep learning optimization module;

[0017] Patent competition factors: ,in The function, which takes into account the number of patents and their influence, is also adjusted by the optimization strategy of the deep learning optimization module.

[0018] Overall innovation potential score: The weight , , satisfy These weights reflect the relative importance of different factors in the comprehensive innovation potential score and are dynamically adjusted according to the actual situation.

[0019] Preferably, in the intelligent recommendation module, the set of recommendation schemes ,in For intelligent recommendation functions, It includes general industry knowledge and expert rules for specific fields. This includes adjustments to the new drug development pipeline, intellectual property protection strategies, partner recommendations, and marketing strategies; the intelligent recommendation function... The deep learning optimization module performs iterative optimization based on historical recommendation data and feedback data.

[0020] Preferably, in the result feedback module, the feedback factor ,in For actual implementation effect, For the discrepancy between input and return, For policy compliance, For enterprise satisfaction, and All indicators were normalized to The result feedback module will synchronize the feedback information to the blockchain collaboration module for evidence storage.

[0021] Preferably, in the dynamic weight tuning module, in the first... In this evaluation cycle, the weight update formula is:

[0022] ;

[0023] ;

[0024] ;

[0025] in, For the learning rate, and , As a reference threshold for the feedback factor, the updated value ensures... The update strategy of the dynamic weight tuning module is affected by the weight optimization model output by the deep learning optimization module.

[0026] Preferably, in the threshold adaptive module, the threshold update formula is: ,in , It is a threshold correction factor, and New threshold The correction factor of the threshold adaptive module is used to determine the enterprise's innovation level in the next round of intelligent recommendation. The deep learning optimization module dynamically adjusts the system based on the company's historical innovation data and market trends.

[0027] Preferably, in the multi-objective recommendation module, the multi-objective recommendation index... The calculation formula is:

[0028] ;

[0029] in, For the evaluation of investment / return, For time-cycle management assessment, This reflects the nature of patent protection and intellectual property strategy. For the feasibility of external cooperation, and The multi-target recommendation module interacts with the blockchain collaboration module to obtain historical cooperation data from partners in order to optimize... The assessment.

[0030] Preferably, in the cooperation value assessment module, the cooperation value index... The calculation formula is:

[0031] ;

[0032] in, For potential partners in the first Quantitative scoring across multiple dimensions For the corresponding dimension weights, and Combined Recommendation Output The calculation formula is:

[0033] ;

[0034] in, , As a balance coefficient between cooperation value and multi-objective recommendation index, the cooperation plan generated by the cooperation value assessment module is automatically executed through the smart contract of the blockchain collaboration module, and the cooperation process data is stored on the blockchain for evidence.

[0035] Preferably, an iterative cycle is formed through data flow: data acquisition module → deep learning optimization module → data analysis module → innovation potential evaluation module → intelligent recommendation module → result feedback module → dynamic weight adjustment module, threshold adaptive module, multi-objective recommendation module, cooperation value assessment module → intelligent recommendation module; at the same time, the blockchain collaboration module runs through the entire process of data acquisition, result feedback, cooperation assessment and execution, realizing continuous optimization of system performance and recommendation results, as well as data security and automation of cooperation processes.

[0036] This invention provides an innovative service system for pharmaceutical companies, which has the following beneficial effects:

[0037] 1. Deep Analysis and Dynamic Optimization of Multi-Source Data: Utilizing deep learning optimization modules (CNN / LSTM) to perform feature extraction and time-series analysis on multi-source data, accurately uncovering data correlations, optimizing data cleaning rules and innovation potential assessment models, making the calculation of factors such as innovation investment, market performance, and patent competition more aligned with the actual needs of enterprises, and improving assessment accuracy.

[0038] 2. Dynamic weight and threshold adaptive mechanism: The dynamic weight optimization module dynamically adjusts the factor weights in the innovation potential assessment based on enterprise feedback, and the threshold adaptive module corrects the assessment threshold in real time through historical data and market trends, ensuring that the system can adapt to the personalized needs of different enterprises and market changes, and avoid the limitations of static configuration.

[0039] 3. Multi-objective recommendation and accurate assessment of cooperation value: The multi-objective recommendation module generates recommendation schemes by comprehensively considering dimensions such as funding, time, patent protection, and cooperation strategies. Combined with historical data of partners obtained by the blockchain collaboration module, it optimizes the feasibility assessment of external cooperation. The cooperation value assessment module achieves efficient matching and automation of the cooperation process through multi-dimensional quantitative scoring and automatic execution of cooperation schemes by smart contracts, thereby reducing the cost of manual decision-making.

[0040] 4. Data security and process automation: The blockchain collaboration module stores original data, feedback information, and cooperation evaluation results on the blockchain to ensure that the data is tamper-proof; through smart contracts, the cooperation plan is automatically executed, improving process transparency and execution efficiency, and providing secure and efficient technical support for enterprise innovation decisions.

[0041] 5. Iterative Optimization Closed Loop Throughout the Process: Each module forms a closed-loop iteration through data flow (data collection → analysis → evaluation → recommendation → feedback → optimization), continuously optimizing system performance and recommendation results. This enables enterprises to obtain suitable innovation strategies in real time, improve R&D efficiency, reduce innovation risks, and ultimately enhance market competitiveness.

[0042] In summary, this invention, through technological integration and process innovation, addresses the shortcomings of traditional systems in areas such as data processing, dynamic adaptation, multi-target recommendation, and data security, providing pharmaceutical companies with intelligent, automated, secure, and reliable innovative service solutions. Attached Figure Description

[0043] Fig. 1 This is a schematic diagram of a pharmaceutical enterprise innovation service system according to the present invention.

[0044] Fig. 2 This is a schematic diagram of the data analysis module of the present invention.

[0045] Fig. 3 This is a flowchart of the innovation potential evaluation module of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Please see Figs. 1-3 This invention provides an innovation service system for pharmaceutical companies, including a data acquisition module, a data analysis module, an innovation potential evaluation module, an intelligent recommendation module, a result feedback module, a dynamic weight optimization module, a threshold adaptive module, a multi-objective recommendation module, a cooperation value assessment module, as well as a deep learning optimization module and a blockchain collaboration module.

[0048] The data acquisition module is used to obtain multi-source data from internal and external channels and store it as a collection. ,in , This represents internal enterprise data. This module represents external data. The deep learning optimization module is connected to the data acquisition module, data analysis module, innovation potential evaluation module, and intelligent recommendation module. It utilizes convolutional neural networks (CNN) and long short-term memory networks (LSTM) to extract features and perform time-series analysis on the acquired multi-source data, uncovering potential correlations between data points and optimizing the data cleaning rules of the data analysis module, the factor calculation function of the innovation potential evaluation module, and the recommendation model of the intelligent recommendation module. The data analysis module is used to clean and preprocess the acquired data to obtain clean data. and from Extract key indicators and integrate them into a vector. This information is then transmitted to the innovation potential evaluation module; the innovation potential evaluation module is used to combine innovation input factors. Market performance factors Patent competition factors and corresponding weights , , Calculate the comprehensive innovation potential score and will The data is passed to the intelligent recommendation module and the threshold adaptive module; the intelligent recommendation module is used to... Together with a knowledge base, a set of recommendation solutions is generated. The results feedback module is used to collect feedback information from enterprises after implementing the recommended plan, and the feedback factors are obtained after normalization. and will The data is passed to the dynamic weight optimization module, the threshold adaptive module, the multi-objective recommendation module, and the cooperation value evaluation module; the dynamic weight optimization module is used for... The weights in the innovation potential evaluation module are dynamically updated; the threshold adaptive module is used to dynamically update the weights based on the innovation potential evaluation module. Compared with the current threshold The difference is used to dynamically update the threshold. The multi-objective recommendation module is used in... Based on this, and considering the needs of enterprises in terms of funding, time, patent protection, and cooperation strategies, a multi-objective recommendation index is calculated. The system generates comprehensive recommendations; the cooperation value assessment module is used to calculate the cooperation value index for potential partners. And when it comes to the selection of collaborators or joint research and development, combined with Output the optimal cooperation combination; the blockchain collaboration module connects with the data collection module, result feedback module, and cooperation value assessment module, and uses blockchain technology to store the collected raw data, enterprise feedback information, and cooperation assessment results to ensure that the data is tamper-proof; at the same time, the cooperation plan generated by the cooperation value assessment module is automatically executed through smart contracts to realize the automation and transparency of the cooperation process.

[0049] More specifically, such as Fig. 2 As shown, the data analysis module includes a data cleaning and preprocessing unit and a key indicator extraction unit; the data cleaning and preprocessing unit uses transformation functions... The multi-source data acquired from the data acquisition module is processed by deduplication, missing value imputation, outlier removal, or normalization to obtain... ,in, Transformation function for the original dataset The optimization parameters are adjusted by the output of the deep learning optimization module; the key indicator extraction unit extracts from... Extracting R&D investment intensity Market share Number of patents Patent influence Key metrics, integration vector This vector will serve as input for the subsequent innovation potential evaluation module.

[0050] The data analysis module is a crucial link in the pharmaceutical enterprise innovation service system, consisting of a data cleaning and preprocessing unit and a key indicator extraction unit. The data cleaning and preprocessing unit performs deduplication, missing value imputation, outlier removal, and normalization on the raw data from both internal and external sources. The processing rules are controlled by the deep learning optimization module, outputting a high-quality dataset. Key indicator extraction unit from Key indicators such as R&D intensity and market share are extracted and integrated into a vector. This module provides input for the innovation potential evaluation module. Its advantages lie in ensuring data quality, dynamic adaptability, and the ability to output structured data, which forms a crucial foundation for the system's accurate assessment of innovation potential.

[0051] More specifically, such as Fig. 3 As shown, in the innovation potential evaluation module, the system combines innovation input factors. Market performance factors Patent competition factors and corresponding weights , , Calculate the comprehensive innovation potential score ;

[0052] The specific calculation method is as follows:

[0053] Innovation input factors: ,in It is a function defined according to enterprise needs and industry characteristics, and is adjusted by the optimization strategy output by the deep learning optimization module;

[0054] Market performance factors: ,in Similarly, it is a function defined according to enterprise needs and industry characteristics, and is affected by the optimization strategy of the deep learning optimization module;

[0055] Patent competition factors: ,in The function, which takes into account the number of patents and their influence, is also adjusted by the optimization strategy of the deep learning optimization module.

[0056] Overall innovation potential score: The weight , , satisfy These weights reflect the relative importance of different factors in the comprehensive innovation potential score and are dynamically adjusted according to the actual situation.

[0057] The innovation potential evaluation module calculates a company's comprehensive innovation potential score through multi-factor weighted calculation. This module constructs an evaluation system based on three dimensions: innovation input factors (reflecting R&D resource investment), market performance factors (measuring market competitiveness), and patent competition factors (assessing patent value and scale). Each factor is calculated using customized functions. For example, the innovation input factor function is influenced by company needs and industry characteristics; the market performance factor function undergoes a special transformation to highlight non-linear relationships; and the patent competition factor function comprehensively considers the quantity and quality of patents. Finally, a weighted summation formula is used to obtain the comprehensive score, where the total weights are equal to 1 to ensure the score's rationality. All factor functions and weights are controlled by a deep learning optimization module, dynamically adjusted according to company characteristics (e.g., startups or mature companies), industry trends, and historical data, achieving adaptive optimization of the scoring model and accurately reflecting the company's innovation potential.

[0058] More specifically, in the intelligent recommendation module, there is a set of recommendation schemes. ,in For intelligent recommendation functions, It includes general industry knowledge and expert rules for specific fields. This includes adjustments to the new drug development pipeline, intellectual property protection strategies, partner recommendations, and marketing strategies; intelligent recommendation functions. The deep learning optimization module performs iterative optimization based on historical recommendation data and feedback data.

[0059] Input: Innovation potential score ,knowledge base( (Including industry-standard knowledge and expert rules)

[0060] Output: Set of recommended solutions Specifically, it includes:

[0061] Adjustments to the new drug development pipeline (such as prioritization and resource allocation);

[0062] Intellectual property protection strategies (such as patent layout and infringement risk prevention);

[0063] Partner recommendations (such as screening joint R&D or manufacturing partners);

[0064] Adjustments to marketing strategies (such as market positioning and competitor responses).

[0065] Intelligent recommendation function Integrating data-driven and knowledge-driven approaches, combining The system uses quantitative scoring and expert experience rules to generate multi-dimensional recommendation schemes.

[0066] Dynamic optimization: Driven by a deep learning optimization module, it iterative training is performed based on historical recommendation data (such as scheme type and execution frequency) and feedback data (such as Ffeed from the result feedback module) to continuously improve recommendation accuracy (for example, by optimizing function parameters through gradient descent to reduce the deviation between recommended schemes and actual needs).

[0067] More specifically, in the results feedback module, feedback factors ,in For actual implementation effect, For the discrepancy between input and return, For policy compliance, For enterprise satisfaction, and All indicators were normalized to The result feedback module will synchronize the feedback information to the blockchain collaboration module for evidence storage.

[0068] Blockchain collaboration: The feedback information (raw data, calculation process, final factor) is synchronized to the blockchain. The hash value is put on the chain to ensure immutability and traceability, and to provide reliable data input for modules such as dynamic weight optimization and threshold adaptation.

[0069] Closed-loop driven: The feedback factor Ffeed triggers the dynamic weight adjustment module (updating the innovation potential assessment weight) and the multi-objective recommendation module (optimizing the recommendation index calculation), forming a closed-loop iteration of "recommendation-execution-feedback-optimization".

[0070] More specifically, in the dynamic weight tuning module, in the first... In this evaluation cycle, the weight update formula is:

[0071] ;

[0072] ;

[0073] ;

[0074] in, For the learning rate, and , As a reference threshold for the feedback factor, the updated value ensures... The update strategy of the dynamic weight tuning module is affected by the weight optimization model output by the deep learning optimization module.

[0075] Based on feedback factors after the recommendation scheme is implemented Dynamically adjust the innovation input factor in the innovation potential evaluation module. ), market performance factors ( Patent competition factor () The weighting of ) ensures that the evaluation model continuously adapts to the actual needs of enterprises and market changes.

[0076] Core variables:

[0077] (learning rate, ): Control the step size of weight updates to ensure that the adjustment range is appropriate (preset) (It can be dynamically optimized by the deep learning module).

[0078] (Feedback Factor Reference Threshold): Serves as a benchmark for measuring feedback effectiveness (e.g., industry average feedback level). ).

[0079] (Partial derivative): represents the weight The degree and direction of the influence on the feedback factor are calculated using the chain rule;

[0080] Constraints: Ensure after update This is achieved through normalization.

[0081] Feedback-driven adaptive adjustment:

[0082] when When the feedback is better than expected, increase the direction of the partial derivative of the effective weights (e.g., by increasing the weights). );

[0083] when When feedback does not meet expectations, the weights are adjusted in reverse to reduce the impact of inefficient factors.

[0084] Deep learning collaborative optimization:

[0085] Weight optimization model: Trains a deep learning model using historical feedback data and dynamically adjusts the learning rate. The partial derivative calculation method improves the stability and convergence speed of weight updates;

[0086] Personalized adaptation: Differentiated weight configurations are achieved through feedback iteration for different enterprise types.

[0087] The dynamic weight optimization module achieves adaptive evolution of the innovation potential assessment model through a feedback factor-driven gradient update mechanism combined with deep learning optimization strategies. The core advantage of this module lies in transforming enterprise performance into quantifiable weight adjustment signals, overcoming the limitations of traditional static configuration systems. This allows the assessment system to respond in real-time to enterprise characteristics, industry trends, and market feedback, thereby significantly improving the accuracy and effectiveness of innovation strategy recommendations.

[0088] More specifically, in the threshold adaptive module, the threshold update formula is: ,in , It is a threshold correction factor, and New threshold Used to assess a company's innovation level in the next round of intelligent recommendation; a correction factor for the threshold adaptive module. The deep learning optimization module dynamically adjusts the system based on the company's historical innovation data and market trends.

[0089] Based on the company's real-time innovation potential score ( ) and the current threshold ( To mitigate deviations, the threshold for judging innovation levels is dynamically adjusted, ensuring that the evaluation criteria of the intelligent recommendation module are adapted to the actual performance of enterprises and market changes.

[0090] Core variables: The difference between the current innovation potential score and the threshold reflects the real-time changes in the company's innovation level (if...). This indicates that the company's innovation performance is above the current threshold, and vice versa. (Threshold correction factor, ( ): Control the magnitude of threshold adjustment to avoid oversensitivity or lag. For example: At the same time, the adjustment increment is 30% of the difference to ensure smooth threshold updates; if the enterprise's innovation data fluctuates significantly, the deep learning module will automatically reduce the threshold. (For example, reduce it to 0.1) to prevent threshold oscillation.

[0091] Update logic: When (Enhanced innovation level): The threshold is raised to increase the benchmark for the next round of evaluation; when (Declining level of innovation): The threshold was lowered to avoid the evaluation criteria becoming detached from the actual situation of enterprises.

[0092] Deep learning-driven correction factor optimization: The deep learning optimization module dynamically adjusts the settings based on the following data:

[0093] Enterprise historical innovation data: Analysis of the past 12 months The fluctuation range (e.g., standard deviation) should be reduced if the fluctuation is large. With a stable threshold;

[0094] Market Trends: By capturing external data such as industry reports and policy changes, if the overall market innovation level improves (e.g., accelerated new drug approvals), the potential for growth will increase. With a rapid response;

[0095] Example: A pharmaceutical company's increased R&D investment led to... If the value exceeds the threshold for three consecutive months, the system will automatically... The value was increased from 0.2 to 0.4, which accelerated the convergence of the threshold to a higher level.

[0096] Avoid the limitations of static thresholds:

[0097] Traditional systems preset fixed thresholds (such as the industry median), which cannot reflect individual differences or dynamic progress of enterprises (such as when the innovation capabilities of startups improve rapidly, fixed thresholds lead to conservative recommendation strategies).

[0098] This module uses real-time deviation calculation and adaptive correction to make the threshold a "dynamic benchmark," for example:

[0099] Company A focuses on research and development. The threshold continues to rise, increasing year by year, and the recommended approach focuses on optimizing innovation investment.

[0100] Company B is in the market expansion phase. Large fluctuations, threshold through small amplitude Smooth adjustments to avoid frequent changes in recommended strategies.

[0101] Accurate assessment of innovation level: The threshold serves as the judgment benchmark for the intelligent recommendation module, and its dynamic adjustment better reflects the current state of the enterprise. For example:

[0102] when When the score is raised to 90% of the company's highest historical score, the recommended approach will focus on strategies to break through bottlenecks (such as joint R&D and high-risk innovation projects).

[0103] If the threshold is slightly adjusted due to an overall market downturn, the recommended approach will shift towards risk control (such as optimizing existing pipelines and strengthening patent defenses).

[0104] It works in conjunction with the innovation potential evaluation module and the intelligent recommendation module to form a closed loop of "score calculation → threshold adaptation → strategy recommendation → effect feedback → threshold re-optimization". For example, after the recommended solution is implemented, if the feedback factor... The display shows improved innovation efficiency, and the threshold adaptive module is adjusted accordingly. This drives the system to continuously uncover higher-level innovation needs. The threshold adaptive module solves the problem that traditional static thresholds cannot adapt to dynamic changes in enterprises through a bias-driven dynamic correction mechanism and deep learning-optimized correction factors. Its core advantage lies in allowing the innovation level assessment benchmark to evolve in real time with enterprise performance and market trends, ensuring that the judgment criteria of the intelligent recommendation module are always accurate and effective, ultimately improving the matching degree between innovation strategies and actual enterprise needs, and facilitating efficient innovation decision-making.

[0105] More specifically, in the multi-objective recommendation module, the multi-objective recommendation index... The calculation formula is:

[0106] ;

[0107] in, For the evaluation of investment / return, For time-cycle management assessment, This reflects the nature of patent protection and intellectual property strategy. For the feasibility of external cooperation, and The multi-objective recommendation module interacts with the blockchain collaboration module to obtain historical cooperation data from partners in order to optimize... The assessment.

[0108] Objective: Based on enterprise innovation potential scoring ( By taking into account multiple dimensions such as funding, time, patent protection, and external cooperation, it generates personalized recommendation solutions that take into account multiple objectives, thus overcoming the limitations of traditional single-dimensional recommendation systems.

[0109] Multi-objective recommendation index It is obtained by weighted summation of the four dimensions:

[0110] ;

[0111] Evaluation dimensions:

[0112] Investment / Return Assessment ( ): Measures the degree of matching between R&D investment and expected returns (e.g., ROI normalized score, range 0,1);

[0113] Time cycle management assessment ( ): Evaluate project cycle efficiency (e.g., whether the R&D cycle meets industry standards; 1 point is awarded for early completion, and points are deducted proportionally for exceeding the deadline).

[0114] Patent protection and intellectual property layout performance ( ): Based on a comprehensive score (expert scoring or algorithmic quantification) of patent coverage technology points, geographical distribution, infringement risk, etc.

[0115] Feasibility of external cooperation ( ): Analyze the technological complementarity and historical cooperation records of potential partners (such as on-time delivery rate and willingness to share intellectual property);

[0116] Weighting: Weight satisfy It can be dynamically adjusted according to the company's strategy (e.g., for companies with tight funds, increasing...). Collaboration-driven enterprises focus on ).

[0117] Blockchain Collaborative Optimization External Cooperation Assessment:

[0118] Data interaction: By connecting with the blockchain collaboration module, we can obtain historical cooperation data from our partners (such as the success rate of R&D of past cooperative projects, records of intellectual property disputes, transparency of financial transactions, etc.). This data has been stored on the blockchain to ensure that it is tamper-proof and reliable.

[0119] Evaluation and optimization:

[0120] Based on blockchain data, machine learning models (such as logistic regression or neural networks) are used to analyze... To quantify, for example:

[0121] Partner A has a historical on-time delivery rate of 90% and a patent sharing rate of 80% for collaborative projects. ;

[0122] Partner B has a history of intellectual property disputes. Automatically lowered by 0.3 points;

[0123] This avoids the problems of traditional assessments relying on subjective judgment or incomplete data, and improves the objectivity and credibility of cooperation feasibility assessments.

[0124] Technical advantages and application scenarios:

[0125] Multi-dimensional balanced recommendation: This addresses the shortcomings of traditional systems that focus only on a single objective (such as optimizing only R&D investment or only considering market share).

[0126] When generating solutions for pharmaceutical startups, you can set up... (Prioritize capital efficiency) + (Feasibility of cooperation), taking into account both cost control and integration of external resources;

[0127] For established pharmaceutical companies that focus on patent strategy, set up We recommend strategies such as building patent pools and strengthening technological barriers.

[0128] Blockchain-enabled partnership evaluation: Quickly screen high-value partners using on-chain data (e.g., from 100 potential partners, based on...). Automatically sort the top 5 results, reducing manual due diligence costs and lowering cooperation risks (such as avoiding historical defaulters).

[0129] Dynamic weighting adapts to strategic needs: Enterprises can flexibly respond to strategic changes by adjusting weights (such as temporarily increasing weights when entering a new therapeutic area). (To shorten the R&D cycle and seize market opportunities).

[0130] More specifically, in the cooperation value assessment module, the cooperation value index... The calculation formula is:

[0131] ;

[0132] in, For potential partners in the first Quantitative scoring across multiple dimensions For the corresponding dimension weights, and Combined Recommendation Output The calculation formula is:

[0133] ;

[0134] in, , To balance the cooperation value with the multi-objective recommendation index, the cooperation plan generated by the cooperation value assessment module is automatically executed through the smart contract of the blockchain collaboration module, and the cooperation process data is stored on the blockchain for evidence.

[0135] Objective: To evaluate the comprehensive value of potential partners through multi-dimensional quantitative scoring and intelligent algorithms, generate the optimal cooperation combination by combining multi-objective recommendation indices, and automate and make the cooperation process transparent through blockchain.

[0136] Basic formula:

[0137] ;

[0138] Input parameters:

[0139] Potential partners in the first Quantitative scoring for each dimension (score range needs to be normalized).

[0140] Weights for each dimension, and .

[0141] Evaluation Dimension Examples:

[0142] Technical dimensions: R&D capability matching degree, core patent coverage;

[0143] Resource dimensions: production capacity and supply chain stability;

[0144] Compliance dimensions: GMP compliance record, intellectual property litigation history;

[0145] Strategic dimension: Market regional complementarity and R&D pipeline synergy potential.

[0146] Optimal combination generation:

[0147] ;

[0148] Balance coefficient: , These are the weights of cooperation value and multi-objective recommendation index, respectively. .

[0149] Application scenario: Risk-averse enterprises: Increase (like Efficiency-driven enterprises: Increase (like ); Provides assessments of internal business needs. Supplement the external value assessment of the partner.

[0150] Automatic execution of smart contracts: After the optimal partner is determined, a smart contract is automatically generated, triggering the following processes: signing of the cooperation agreement, phased disbursement of funds, and on-chain synchronization of progress; the cooperation value assessment module upgrades the cooperation decision-making process through a multi-dimensional quantitative assessment model and blockchain smart contracts.

[0151] Its core innovations lie in: building a scientific cooperation value assessment system; leveraging blockchain technology to improve efficiency and reduce risks; and this module helps pharmaceutical companies efficiently select partners in scenarios such as joint R&D and production outsourcing, thereby promoting the optimal allocation of innovation resources.

[0152] More specifically, an iterative cycle is formed through data flow: data acquisition module → deep learning optimization module → data analysis module → innovation potential evaluation module → intelligent recommendation module → result feedback module → dynamic weight adjustment module, threshold adaptive module, multi-objective recommendation module, cooperation value assessment module → intelligent recommendation module; at the same time, the blockchain collaboration module runs through the entire process of data acquisition, result feedback, cooperation assessment and execution, realizing continuous optimization of system performance and recommendation results, as well as data security and automation of cooperation processes.

[0153] Core Logic: Through data interaction between modules, a closed loop of "collection-analysis-evaluation-recommendation-feedback-optimization" is formed, enabling continuous evolution of system performance and recommendation results. Iterative cycles, driven by data from inter-module closed-loop optimization, achieve precise and adaptive innovative services, addressing the inefficiency of static configuration in traditional systems. Blockchain collaboration, through end-to-end notarization and automated execution, ensures data security, improves collaboration efficiency, and addresses the pain points of data tampering risks and high costs of manual intervention. The combination of these two elements makes the system an innovative service platform integrating "intelligent evaluation, dynamic optimization, and secure execution," providing pharmaceutical companies with technological support for sustainable evolution.

[0154] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A pharmaceutical enterprise innovation service system, characterized in that, It includes modules for data acquisition, data analysis, innovation potential evaluation, intelligent recommendation, results feedback, dynamic weight optimization, threshold adaptation, multi-objective recommendation, collaborative value assessment, deep learning optimization, and blockchain collaboration. The data acquisition module is used to acquire multi-source data from internal and external channels and store it as a collection. ,in , This represents internal enterprise data. Represents external data; The deep learning optimization module is connected to the data acquisition module, data analysis module, innovation potential evaluation module, and intelligent recommendation module. It uses convolutional neural networks and long short-term memory networks to perform feature extraction and time series analysis on the acquired multi-source data to uncover potential correlations between data. The data analysis module is used to clean and preprocess the collected data to obtain clean data. and from Extract key indicators and integrate them into a vector. This information is then transmitted to the innovation potential evaluation module. The innovation potential evaluation module is used to combine innovation input factors. Market performance factors Patent competition factors and corresponding weights , , Calculate the comprehensive innovation potential score And The data is then passed to the intelligent recommendation module and the threshold adaptive module. The intelligent recommendation module is used for... Together with a knowledge base, a set of recommendation solutions is generated. ; The result feedback module is used to collect feedback information from enterprises after implementing the recommended scheme, and obtain feedback factors after normalization. and will The data is then passed to the dynamic weight optimization module, the threshold adaptive module, the multi-objective recommendation module, and the cooperation value assessment module. The dynamic weight adjustment module is used to base on The weights in the innovation potential evaluation module are dynamically updated. The threshold adaptive module is used to... Compared with the current threshold The difference, dynamically update the threshold. ; The multi-objective recommendation module is used to... Based on this, and considering the needs of enterprises in terms of funding, time, patent protection, and cooperation strategies, a multi-objective recommendation index is calculated. And generate a comprehensive recommendation scheme; The cooperation value assessment module is used to calculate a cooperation value index for potential partners. And when it comes to the selection of collaborators or joint research and development, combined with Output the optimal combination of cooperation; The blockchain collaboration module is connected to the data acquisition module, the result feedback module, and the cooperation value assessment module. The data analysis module includes a data cleaning and preprocessing unit and a key indicator extraction unit. The data cleaning and preprocessing unit uses a transformation function. The multi-source data acquired from the data acquisition module is processed by deduplication, missing value imputation, outlier removal, or normalization to obtain... ,in, For the original dataset, the transformation function Adjusted by the optimization parameters output by the deep learning optimization module; The key indicator extraction unit Extracting R&D investment intensity Market share Number of patents Patent influence Key metrics, integration vector This vector will serve as the input for the subsequent innovation potential evaluation module; In the innovation potential evaluation module, the system combines innovation input factors. Market performance factors Patent competition factors and corresponding weights , , Calculate the comprehensive innovation potential score ; The specific calculation method is as follows: Innovation input factors: ,in It is a function defined according to enterprise needs and industry characteristics, and is adjusted by the optimization strategy output by the deep learning optimization module; Market performance factors: ,in Similarly, it is a function defined according to enterprise needs and industry characteristics, and is affected by the optimization strategy of the deep learning optimization module; Patent competition factors: ,in The function, which takes into account the number of patents and their influence, is also adjusted by the optimization strategy of the deep learning optimization module. Overall innovation potential score: The weight , , satisfy These weights reflect the relative importance of different factors in the comprehensive innovation potential score and are dynamically adjusted according to the actual situation. In the dynamic weight optimization module, at the first In this evaluation cycle, the weight update formula is: ； ； ； in, For the learning rate, and , As a reference threshold for the feedback factor, the updated value ensures... The update strategy of the dynamic weight tuning module is affected by the weight optimization model output by the deep learning optimization module.

2. The pharmaceutical enterprise innovation service system according to claim 1, characterized in that: The intelligent recommendation module contains a set of recommendation schemes. ,in For intelligent recommendation functions, It includes general industry knowledge and expert rules for specific fields. This includes adjustments to the new drug development pipeline, intellectual property protection strategies, partner recommendations, and marketing strategies; the intelligent recommendation function... The deep learning optimization module performs iterative optimization based on historical recommendation data and feedback data.

3. The pharmaceutical enterprise innovation service system according to claim 2, characterized in that: In the result feedback module, the feedback factor ,in For actual implementation effect, For the discrepancy between input and return, For policy compliance, For enterprise satisfaction, and All indicators were normalized to The result feedback module will synchronize the feedback information to the blockchain collaboration module for evidence storage.

4. The pharmaceutical enterprise innovation service system according to claim 3, characterized in that: In the threshold adaptive module, the threshold update formula is: ,in , It is a threshold correction factor, and New threshold The correction factor of the threshold adaptive module is used to determine the enterprise's innovation level in the next round of intelligent recommendation. The deep learning optimization module dynamically adjusts the system based on the company's historical innovation data and market trends.

5. The pharmaceutical enterprise innovation service system according to claim 4, characterized in that: In the multi-objective recommendation module, the multi-objective recommendation index The calculation formula is: ； in, For the evaluation of investment / return, For time-cycle management assessment, This reflects the nature of patent protection and intellectual property strategy. For the feasibility of external cooperation, and The multi-target recommendation module interacts with the blockchain collaboration module to obtain historical cooperation data from partners in order to optimize... The assessment.

6. The pharmaceutical enterprise innovation service system according to claim 5, characterized in that: In the cooperation value assessment module, the cooperation value index The calculation formula is: ； in, For potential partners in the first Quantitative scoring across multiple dimensions For the corresponding dimension's weight, and Combined Recommendation Output The calculation formula is: ； in, , As a balance coefficient between cooperation value and multi-objective recommendation index, the cooperation plan generated by the cooperation value assessment module is automatically executed through the smart contract of the blockchain collaboration module, and the cooperation process data is stored on the blockchain for evidence.

7. The pharmaceutical enterprise innovation service system according to claim 6, characterized in that: An iterative cycle is formed through data flow: data acquisition module → deep learning optimization module → data analysis module → innovation potential evaluation module → intelligent recommendation module → result feedback module → dynamic weight adjustment module, threshold adaptive module, multi-objective recommendation module, cooperation value assessment module → intelligent recommendation module; at the same time, the blockchain collaboration module runs through the entire process of data acquisition, result feedback, cooperation assessment and execution, realizing continuous optimization of system performance and recommendation results, as well as data security and automation of cooperation processes.

Images