Healthcare Data Storage Issues? Yes, AI Can Help with That Too

The explosion of data in all industries, especially healthcare, is well documented. So, it’s not surprising that healthcare organizations are looking to better manage data storage.  

The imminent explosion of artificial intelligence (AI) is also well documented. So, it’s not surprising that AI is emerging as a potential data storage management tool. In fact, the AI-powered data storage market is expected to hit $10.4 billion by 2025, growing at a compound annual growth rate of 34.4 percent, according to a report by MarketsandMarkets.  

“In the United States, AI is being used for data storage because organizations now understand the significance of the data they own and they understand the reason why they need to store it better and manage it better,” says Ayomide Owoyemi, PhD, research assistant in the department of biomedical and health information sciences at the University of Illinois, Chicago. “And, because these organizations are using AI for other purposes, they are generating even more data.”  

Indeed, AI can provide much more than traditional storage management systems as healthcare organizations strive to handle increasing volumes of complex data. According to a column in Insights for Professionals, AI-enabled storage solutions leverage machine learning algorithms and advanced analytics to analyze patterns in data usage, optimize storage resources, automate routine tasks, and eliminate human errors, resulting in improved efficiency and cost savings.  

Future Forecasting 

Perhaps most importantly, AI can help health information (HI) professionals predict data storage needs for their organizations. 

“Data storage is expensive, and organizations need to understand their data storage needs in anticipation of data creation. It's not something you do in retrospect,” Owoyemi says.  

In addition, AI can help to determine data accessibility requirements.  

“The imaging data, the CT scans, the MRIs, even the data coming from individual patients, it is all just growing exponentially. Having all that data accessible within milliseconds of time is just too cost prohibitive,” says Andrew Boyd, MD, associate professor in the department of biomedical and health information sciences at the University of Illinois, Chicago. With AI, HI professionals can determine what data needs to be immediately accessible and what data needs to be available in a day, week, or month.  

AI’s Advantages 

In addition to predicting data storage and accessibility needs, Harshit Shah, CTO at Kyruus Health, a healthcare technology company in Boston, says that healthcare organizations can leverage AI for:  

Predictive maintenance. Advanced data collection, analysis, and predictive modeling techniques can help to predict maintenance needs. By predicting equipment failures before they occur, AI can help reduce downtime, optimize maintenance schedules, extend equipment lifespan, improve patient safety, and ensure regulatory compliance.  

Performance optimization. AI can ensure that storage systems operate efficiently, quickly, and reliably, which is crucial for handling the large volumes of sensitive data typical in healthcare environments. In addition, AI can help with dynamic resource allocations based on changing traffic patterns, by auto-tiering to keep frequently used datasets in cache versus others in cost-effective storage buckets. AI can also help to assess in-depth root cause analysis on performance issues and generate synthetic data/traffic to simulate real world patterns to validate the performance of a system under load. 

Regulatory compliance. By monitoring the data access and maintaining forensic trails — that can be alerted upon for anomalous patterns — AI can help healthcare systems stay compliant with regulatory requirements such as the Health Information Technology for Economic and Clinical Health (HITECH) Act. 

Advanced data analysis.  AI can analyze large datasets and recognize patterns. This can ultimately improve care access trends, patient care journeys, and provider workflows. 

Data protection. Healthcare data contains highly sensitive information including personally identifiable information/protected health information (PII/PHI) data. Protecting this data from breaches and unauthorized access is paramount. AI can be leveraged to meet the strict regulations around HIPAA and General Data Protection Regulation (GDPR) compliance — as well as local laws that govern data privacy.    

Leveraging AI for these purposes is particularly useful, as healthcare organizations face unique challenges. 

“Data scalability and sensitivity significantly affect healthcare data storage due to the industry's unique demands for privacy, compliance, data integrity, and real-time access. While these concerns are present in other industries, they are particularly heightened in healthcare because of the direct impact on patient safety and the stringent regulatory environment,” Shah says. 

AI can also help alleviate some of the manual data management tasks for HI professionals.  

“How do you label the data? How do you arrange the data? A lot of these things are done manually right now, and health information professionals manage these processes,” Owoyemi says. “There's a chance to use AI to either make these processes more effective or, in some cases, if you can develop a very, very good algorithm to reduce the need for human resource in that area.”  

By leveraging AI in these ways, healthcare organizations can reap many benefits. To start, AI can help reduce data storage costs. Simply understanding data storage needs in advance and increasing data retrieval efficiency could help organizations reduce data storage and human resource costs, according to Owoyemi.  

In addition, using AI to better protect data can also reduce costs. “The healthcare industry recently has had a couple of significant cybersecurity attacks that are very expensive to recover from. Health information leaders could leverage AI to predict and then prevent attacks,” Owoyemi says.  

More specifically, AI-based threat detection can analyze extensive data, pinpointing abnormal behaviors and malicious activities. It can detect anomalies, isolate compromised machines, and promptly halt attacks. Through real-time continuous monitoring, AI can identify anomalies and intrusions, detect fake users, mitigate attacks, and quarantine infected devices. Also, AI can continuously evaluate the trustworthiness of devices, users, and applications, providing immediate responses to breaches and fraudulent activities, Owoyemi says. 

AI’s Flipside

While AI could enhance data storage, Shah pointed out that HI leaders need to consider a variety of issues that are associated with any AI application.  

Perhaps most importantly, leaders need to understand that AI systems require a large amount of high-quality data to get trained and function effectively. Incomplete data can lead to incorrect predictions or decisions. In addition, malicious actors could potentially exploit the AI systems to gain access to underlying datasets.  And finally, it’s easy to underestimate time and efforts for the initial setup and integration of AI systems, as well as the resource intensive nature of on-going maintenance, according to Shah.  

In addition to these common AI concerns, the government could pass regulations that change data storage needs for healthcare organizations and there would be no way AI could detect such occurrences. Similarly, insurers could change data requirements as well. 

“AI is not a magic wand. These things are very dynamic, and AI doesn't just automatically account for all of this,” Boyd says. “Hopefully, the AI is sophisticated enough to provide a level of certainty, but it can necessarily predict when there will be huge policy changes that affect storage needs. I know of no one who can predict these macro trends. Our health system is just so complex and there are so many players that can provide impact on data trends.”   

Leaders also need to be aware of challenges associated with data drift, which can have an impact on AI’s performance. “If the performance of AI on data storage is trained off 2022 data and it's 2024, some of these changes might not be accounted for. So, the benefit of AI might not be there,” Boyd says.  

AI’s error rates also come into play, he adds. The error rate of AI tools that are acceptable to hospitals is a question. Some healthcare organizations might be willing to live with a 10 percent error rate. Others will not and will still have HI professionals reading the entire document. “So ultimately, no tool is perfect, but humans aren't perfect either,” Boyd says.  

Health leaders then need to determine if it is best to purchase a third-party AI-empowered data storage solution or build one in-house. “Building in-house could take longer because you might not have the necessary human resource to make the happen very quickly,” Owoyemi says. “So, buying a solution has advantages, but then there are challenges. For instance, there’s a security risk with a third-party having access to your data.” 

Perhaps most importantly, though, before buying or building an AI solution for data storage, HI leaders should evaluate AI’s overall value. To do this, leaders should weigh all of AI’s performance advantages against the risks and limitations associated with the emerging technology. Doing so can make it possible to determine exactly how much AI can reduce costs, enhance data security, and improve efficiency.  

John McCormack is a Riverside, IL-based freelance writer covering healthcare information technology, policy, and clinical care issues.