Here’s a startling statistic: health records are worth 40 times more than credit card data on the black market.
With hackers increasingly targeting this lucrative data, it’s no surprise that cybersecurity is top-of-mind for healthcare executives.
Artificial intelligence and machine learning are already helping hospitals all over the world ease their administrative burdens and improve patient health outcomes. Now, AI is being used in cybersecurity to help augment human efforts and ensure the safety of patient data.
Of course, AI implementation in healthcare security still faces several hurdles, and malicious AI apps, which work against security systems, are a growing threat. In order to ensure data security and patient safety, maintaining focus on addressing these issues will become essential.
Here are four key ways AI can combat healthcare data security challenges and help protect hospital and medical device networks — today and as we head into the future of healthcare.
1. Identify new malware threats with machine learning
Machine learning apps can identify emerging threats using predictive algorithms.
For decades, IT security has relied on detecting the signatures of specific malicious programs. The problem with this approach is that it’s reactive; a breach must occur before such an attack can be prevented in the future.
Nearly 1 million new malware threats appear every day, each with a new digital fingerprint. This evolution is outpacing traditional security measures, skewing the advantage in attackers’ favor. After all, a virus only has to work once — security systems have to work every time.
Against this onslaught, it’s clear why hospitals face frequent security breaches, even with advanced defenses in place. On average, more than fifteen thousand patient records are compromised every day. In 2017, this added up to a grand total of over five million patient records breached over the course of the year.
The costs of these breaches are massive: on average, a single compromised record costs hospitals $380. Using simple math, this means that U.S. hospital systems are losing over $5 million per day from data breaches alone, a figure which doesn’t take into account damage to hospital reputability. In one study, 50% of respondents indicated that a breach of their data would lower their trust in their providers or prompt them to change providers.
The human cost of stolen patient data is more difficult to quantify, but even more concerning. Health data is not protected from fraud in the same way credit reports are, making medical identity theft hard to recover from. Furthermore, if data breaches result in false patient records, patients could risk severe health consequences.
Machine learning apps can help prevent these breaches by proactively searching for previously unknown malware signatures. Using historical data, these AI applications learn to recognize malware patterns even when the destructive program is not previously known.
Yet, there are still barriers to the full-scale implementation of this technology in healthcare IT. Most notably, automating the process will require access to massive healthcare data sets, which are often protected by HIPPA regulations.
There is also a sinister arms-race in progress. As cybercriminals ramp up their efforts to target increasingly lucrative medical data, they can design malicious AI apps to systematically target vulnerabilities in hospital security systems.
Significant developments in AI-supported security programs have already been made since 2017, however. As a result, AI has already proven effective at boosting security efforts in other industries.
Continuing to develop apps that enhance hospital systems’ capacity to proactively respond to threats will be increasingly crucial as cybercriminals continue to refine their attacks.
Takeaway: Cybersecurity requires visibility; you cannot protect yourself from something you don’t know exists. Machine learning represents one of the only ways to make previously unknown threats visible. Investing in this technology is key to securing healthcare data against novel, widespread and damaging cyber attacks.
2. Identify and respond to breaches using behavioral modeling
If a security breach does occur, AI helps identify and isolate these threats faster than traditional security measures.
AI automatically and continuously monitors behavior within a network by flagging anomalies as they occur.
Behavioral indicators are powerful tools for identifying malware. For example, suspicious events — say, a staff member’s account suddenly accessing 5,000 patient files at 4 a.m. — create behavioral patterns which AI can learn to identify.
Once the threat is detected, AI technology can either forward the case to central human oversight or be designed to take autonomous action to reduce the impact of the breach, should certain criteria be met.
For example, AI-powered automation can defensively segment traffic to automatically isolate sensitive data based on specific security protocols. This helps ensure that compromised devices can’t spread the infection over the entirety of the healthcare network.
Takeaway: AI can help to quickly identify anomalous behavior in networks and act to isolate it. Implementing AI technology augments human oversight and extends existing security resources, helping your organization stay ahead of breaches and minimize any damage.
3. Protect medical devices from attack
Smart medical devices represent serious threats to patient safety if left unprotected from remote breaches. AI can help address the most challenging barriers to securing these devices.
There are currently 3.7 million connected medical devices in use in the U.S. This figure works out to about 10-15 devices per bed in the average American hospital.
Defibrillators, pacemakers, insulin pumps and other medical electronics offer huge benefits for patient health. But according to recent research, many are also vulnerable to attack. Hundreds of thousands of these personal devices are publicly discoverable, and more than 3% currently use outdated operating systems that no longer receive security updates.
Medical electronics — including implants and larger equipment such as scanners and MRI machines — are typically designed by third-party suppliers who use diverse operating systems and proprietary codes. This makes it difficult for hospital systems to assess equipment for security flaws.
Hospitals are responsible for the integrity of the devices they connect to their patients, but they don’t have full control over security measures to keep patients safe while using them. Demonstrated vulnerabilities in many of these devices could lead to compromised data or actually influence their performance. The extent of this risk is difficult to fully calculate.
If not properly safeguarded, it’s likely that future cyber-attackers will be able to exploit vulnerabilities in these devices to target patients, their data, and even the hospital systems to which they are connected. Fake medical devices could be inserted within hospital networks, impersonating devices such as CT scanners and MRI machines, and be used to access the network to shut it down or even hold it for ransom.
Clearly, threat detection in medical electronic devices is a key area for further research. AI can help with both data encryption and monitoring for malware, particularly in the automatic identification of threats, without being dependent on manufacturers to reveal vulnerabilities.
Takeaway: Hospitals need to work diligently to protect themselves and their patients from medical device vulnerabilities, and soon. Artificial intelligence is a virtually untapped resource for the evaluation of data coming from connected devices, as well as the management of access these devices have to a hospital’s network.
4. Extend human resources and address security shortages
The sheer quantity of potential threats can overextend security staff, reducing their efficacy and contributing to burnout.
There is a shortage of qualified IT security professionals. According to CIO Magazine, more than 30% of healthcare managers report having to postpone or shrink an IT project due to staffing issues. And even when IT professionals are available, hiring them is expensive and represent a significant drain on hospital resources.
Many existing IT security measures only work if there is someone actively monitoring the system at the time of the attack. These employees must meticulously and continuously check access logs to prevent and react to unauthorized activity.
Exactly how much information needs to be monitored?
Some organizations see over 200,000 possible security events every day. The sheer quantity of potential threats can overextend security staff, reducing their efficacy and contributing to burnout.
Still, hospital systems under attack need to rapidly triage potential threats.
In 2016, IBM found that the average organization’s security systems waste over $1.3 million — and more than 21,000 hours of work — on false positives alone every year.
Artificial intelligence solutions allow hospitals to process massive datasets rapidly and efficiently, freeing professional to focus their intuition and experience on the highest-priority potential threats.
AI apps such as IBM’s Watson are already speeding up routine security assessments, reducing response times and the time spent chasing down false positives. They can also speed up data-processing on historical cybercrime data and provide analysts with recommendations based on their analysis of extensive security data, enabling security staff to focus their efforts on improving systems and addressing the most urgent and complicated tasks.
The net result is faster decision-making and reduced pressure on staff. Outside healthcare, companies using AI apps to extend the reach of security are already seeing a big impact on response times: IBM research found that average investigation times have decreased from 60 minutes to only one minute for early adopters.
Takeaway: AI solutions ease the pressure of IT security staffing gaps and extend the capacity of existing security teams. Automating IT security tasks is an immediate way to both reduce financial liability from cyber attacks and use human resources more efficiently.
Cyberattacks cost hospitals $6.2 billion every year and represent a serious danger to patients and administrators alike.
In response, nearly 90% of healthcare executives already report using traditional anti-virus software, but this is no longer enough to ensure security.
AI-driven security solutions aren’t a magic bullet, but they will become a critical tool for executives to tap in order to create a robust, modern security system. Despite certain barriers to implementation, AI technology currently on the market is helping hospital cybersecurity teams:
- – Identify new threats
- – Respond to and isolate data breaches
- – Protect connected medical devices
- – Extend security resources
In the coming years, implementing AI-supported cyber defenses will become essential if the healthcare industry is to proactively protect and respond to cyber threats, keeping patients and their data safe.