The private and secure sharing of data may sound like a noble undertaking and in many cases, it is, but the primary motivation for organizations and government agencies to exchange sensitive information is to unlock valuable insights.
The international Organization for Economic Co-operation (OECD) has estimated that data sharing and increased access are capable of producing up to 1.5 percent of a country’s GDP with respect to public-sector data, and between 1 and 2.5 percent of GDP for private-sector data.
The value created by secured data collaboration comes in the forms of greater efficiency and innovation. Medical researchers can share sensitive data to better treat diseases like tuberculosis. Carmakers developing self-driving vehicles can share data to make their automated systems safer. Tech companies can share GPS data with government agencies to help design better traffic flows.
Secure data collaborations are often used to make artificial intelligence systems less biased, more accurate, and more functional. An AI system is only as good as the data used to train it, so if the system only has access to a limited dataset, that severely limits its capabilities. Secure data collaborations can also break down data silos so that larger sets can be aggregated, leading to higher-quality AI models.
In data collaborations that involve sensitive data, maintaining privacy is a major concern. Healthcare organizations looking to share medical information need to ensure that the identities of patients are not revealed. Financial companies looking to create better products using customer data need to protect the financial information of individual customers.
Privacy-enhancing technologies have been developed to address this concern. Through methods like differential privacy and the TripleBlind Solution’s innovative technology, organizations can address privacy concerns while producing valuable insights.
Consider these examples of secure data collaboration unlocking valuable solutions for stakeholders.
Fighting Tuberculosis in India
According to the Centers for Disease Control and Prevention, 1.7 billion people around the world were infected with tuberculosis in 2018, equal to about 23 percent of the global population.
To battle this crisis in India, a secure data collaboration between mobile provider Airtel and the World Health Organization’s “Be He@lty, Be Mobile” initiative was able to develop a proof of concept method for identifying areas at risk for increasing tuberculosis cases.
Since tuberculosis is spread through recurring proximity among individuals, location data of mobile phones can be used to assess the risk for increased spread of the disease. The data partnership between Airtel and the WHO used anonymized data from the company’s mobile network to identify patterns of population movements, such as commuting patterns and other daily routines. The precision, scale, and immediacy of mobile data provided by Airtel allowed researchers to identify at-risk areas. Analysis of the mobile data revealed movement patterns are a more reliable indicator of tuberculosis incidence than proximity to tuberculosis hot spots.
Now that these patterns and insights have been unlocked, the Indian government can take better public health steps related to the prevention and treatment of tuberculosis. More broadly, this collaboration revealed how the GPS data of individuals can be used for public benefit while still maintaining privacy.
WorkFusion is a tech company that offers business tools capable of automating standard, time-consuming tasks — such as invoice processing. The company’s platform uses character recognition technology to read documents and extract relevant information from them. The tool was developed to process invoices, but it is capable of other similar applications that can be found in just about every industry.
One of the main challenges in opening up the system to have broad appeal is the fact that different industries use different types of documents and document formats. An invoice in the supply chain and an invoice in the medical industry might serve the same basic purpose but appear very different from each other. For the WorkFusion platform to be capable of processing invoices from many different industries, it must be able to recognize key information regardless of document format.
To achieve this, the platform’s AI model must be trained using the types of documents it will be processing. Unfortunately, companies do not freely publish internal documents and so the WorkFusion AI model had to be trained on a per-customer basis. Lasting as long as six months, the necessary training period did not generate revenue for WorkFusion, and customers had to disclose months’ worth of internal documents before being able to realize any benefit.
WorkFusion addressed this “cold start” problem by combining deep learning models with privacy-enhancing technology. First, the company improved its existing machine learning capabilities with a new deep learning model. The company then developed the capability to extract key insights from a broad range of its customers through the generation of synthetic invoice data. Finally, the company used differential privacy to aggregate all of its customers’ data, including new customers, while guaranteeing that individual data contributions were kept private.
The new and improved WorkFusion platform is now able to get new customers up and running after just a few days of AI training, reducing the delay on incoming revenue. One of the company’s financial services customers was so impressed with the technology that it significantly expanded its partnership with the company after one year, resulting in a major revenue increase.
How TripleBlind Can Unlock Value in Your Next Data Collaboration
From preventing credit card fraud in the financial services industry to facilitating genetic analysis in healthcare, TripleBlind can ensure the privacy of individuals while allowing data collaborators to unlock the intellectual property value of data.
Delivered via simple API, the TripleBlind Solution can address a wide range of use cases. It allows for true scalability and faster processing of sensitive data, improving the practical use of privacy-preserving technology. Our technology is able to handle all data types and natively supports all major cloud platforms. It compares favorably to other privacy technology like differential privacy and federated learning. Contact us today to learn more.