Protectability of AI diagnostics in health care

Health care innovators continue to develop artificial intelligent (AI) algorithms to aid them in diagnosing and treating patients. However, while these algorithms are very valuable, they are tricky to protect as intellectual property.

Many companies in fields outside health care protect their AI algorithms with trade secrets because they innovate too fast for patents to be worthwhile. They would not be able to detect when competitors are infringing anyway. But in the case of health care, particularly diagnostics, algorithms must be verified by doctors and regulators—so trade secret protection gets complicated, as at least part of the algorithms might need to be disclosed.

Since trade secrets are not an option to fully protect them, companies must develop another way to sufficiently protect their algorithms to justify the cost of clinical trials. As a result, the only option is to patent. Unless the government makes significant changes, AI-based diagnostics will continue to be tricky to patent, since software and diagnostic patent laws are constantly evolving.

But with some legal creativity, there are multiple paths to patent protection for AI-based diagnostics. As opposed to technology that resides in a field, the combination of software and diagnostics can actually be easier to patent because it relies on patent laws originating in both the software and biotech fields.

First, since many diagnostics require sensors, algorithms can be patented if tied to the use of specific sensors, like, for example, pulse oximeters. This is one avenue health care companies typically use to overcome software patentability obstacles. In cases where diagnostics do not require sensors, the patents can be tied to a particular treatment once there is a diagnosis—one of the classic linchpins for patenting in the biotech space. In many cases, having both types of patents also provides better protection than would be available for either only software or diagnostics.

Second, the details of algorithms can still be kept as trade secrets. Particularly, the training data and further details can generally be kept secret and even provide a competitive advantage. Trade secret protection, combined with unique patenting strategies, usually provides sufficient protection for AI-based diagnostics. Therefore, health care companies would be wise to develop AI-based diagnostics and precision medicine tools, but carefully craft their IP strategy using patents to fully protect these innovations.

Tags: Health Care