Is This the Silver-Bullet Evidence That Apple Couldn’t Find?

Data-Driven Analysis on Masimo vs Apple

The case has garnered significant attention in the technology and legal communities. At the heart of this dispute lies U.S. Patent No. 10,945,648 issued to Poeze et al. (the '648 patent), which is generally directed to user-worn devices for non-invasively determining physiological parameters, including blood oxygen saturation.

It did not take long for us to appreciate that this is an edge case of patent litigation, complexity-wise. The prosecution history is extensive, as is the volume of prior art, not to mention the numerous judicial opinions on claim construction, validity, and infringement.

Comprehensive analyses of the usual suspects of issues thus appeared to have been arduous, if not difficult, even for the most sophisticated patent professionals, which makes this a great test case for the technology we’ve developed. We leveraged our proprietary technology, fusing a patent knowledge base with the power of AI to (i) find, with high recall and precision, documents material to patentability and infringement and (ii) analyze them programmatically. We learned several things about our system with regard to the case, and would like to share our findings with those interested in our technology and/or the landmark dispute.

Key Issues in the Case

The legal battle between Masimo Corporation and Apple Inc. revolves around Masimo’s allegations that the Apple Watch product line that measures blood oxygen saturation infringe Masimo's patents, including the '648 patent. Claims of the ‘648 patent recite, generally speaking, a user-worn device configured to non-invasively measure physiological parameters such as oxygen saturation levels, a feature central to health monitoring applications in wearable devices.

The claims in contention include claims 8 and 12, which priority date appears to be as early as July 3, 2008 and may be as late as August 3, 2009, depending on the filing date of the parent specifications that can provide sufficient Section 112 support. Together these claims recite, amongst other elements, sets of LEDs that emit light at different wavelengths, photodiodes, and a protrusion with a convex surface made partially from opaque material. These components are configured and work together to measure non-invasively oxygen and oxygen saturation levels.

In assessing the claims asserted in Masimo vs Apple, it was imperative to dissect the language of the patent and assess the scope of its protection. The term "protrusion," for instance, was a focal point of analysis because its construction directly impacted the scope of the claims in a way that was material to both validity and infringement. To obtain a precise and comprehensive understanding of how a POSITA would construe this term, we leveraged our technology to comb through the intrinsic evidence and found instances where the description shed light on structure and function. Under this construction of the term, we then programmatically analyzed both prior art and accused products and assessed their materiality.

Methodology for Deep Diving the ‘648

Most patent analysis typically involves comparing and contrasting a patent claim at issue to some technology of interest. Essential to accurate and compelling analysis is a precise understanding of the claim, especially its scope. Notably, the technology of interest may be described in a patent, printed publication, product specification, research paper, and/or source code. Moreover, the technology of interest can be prior art or an infringing product, depending on priority dates, release dates, publication dates, commit dates, and so forth.

With these things in mind, we adhered to a methodology that leverages a few well known and simple, but powerful hypotheses, which are:

1. Accurate claim construction requires exhaustive and rigorous review of the intrinsic evidence.
2. Reliable assessments of validity and infringement require information retrieval that has both high recall and precision, which means searching for documents in corpora beyond patent databases and indices of websites. (This search beyond the customary brings us closer to the POSITA’s presumed extraordinary knowledge of all documents publicly available in the world.) Also necessary are thorough and rigorous reviews of the documents found to assess relevance (here, whether they are material to validity or infringement).
3. These requirements can be achieved more effectively and more efficiently with Patented.ai’s developing technology.

To test our hypotheses, we used our AI engine to extract a term glossary, which includes amongst others the term “protrusion”, and analyzed everything found in the intrinsic evidence to understand the various constructions for that term that are colorable. Then the system programmatically searched non-patent corpora and, where promising, followed leads found there. Finally, we used our engine to analyze results to assess their relevance.

Our programmatic approach proved time-efficient and exhaustive, providing us with a comprehensive understanding of claim terms like "protrusion" and proving our hypotheses as sound.

Analyzing the Claim Term "Protrusion"

As part of our work to understand the claims at issue, Patented.ai's Claim Analysis engine programmatically extracted relevant disclosures from the intrinsic evidence of the '648 patent. This was an iterative process and ultimately produced a list of descriptions of the structures and functions associated with the term, each accompanied by an indication of confidence level in the accuracy of the description.

Structure: “Protrusion”

Structure	Confidence	Analysis
The protrusion is a physical component that is part of the noninvasive sensor housing design.	High	The text clearly and unambiguously states "an exemplary noninvasive sensor housing including a finger bed protrusion", directly indicating the protrusion is a physical part of the sensor housing.
The protrusion is a physical component with side and top perspective views.	High	The text states "side and top perspective views of an example protrusion", directly specifying the protrusion has side and top views as a 3D object.
The protrusion is a physical component that affects light transmittance.	High	The text states "effects of a protrusion on light transmittance", clearly establishing light transmittance as a function.
The protrusion is a physical component incorporated into the sensor design.	High	The text states "an example sensor incorporating the protrusion", unambiguously indicating the protrusion is incorporated into the sensor.
The protrusion conforms tissue to reduce mean path length.	High	The text states the "protrusion is used to thin out the measurement site" to "increase the SNR", directly linking tissue conformation to reduced path length and increased SNR.
The protrusion must have a contact area that contacts body tissue.	High	The text states the protrusion "includes a measurement site contact area" and that it "can contact body tissue", clearly requiring tissue contact.
The protrusion contains shielding to reduce noise.	Medium	The text states it "can include" shielding as one option but does not mandate shielding.
The protrusion is a physical component having a partially cylindrical shape.	Low	This describes one embodiment but does not limit the protrusion only to a partially cylindrical shape. Other sections indicate it can have other shapes.
The protrusion is a physical component with a front elevation view.	Low	This illustrates one embodiment but does not strictly require the protrusion to have a front view.
The protrusion has a convex bump shape.	Low	While this describes one protrusion embodiment shape, other sections indicate other shapes are possible.
The protrusion includes openings on its external surface.	Low	The text states "external surface of the protrusion can include one or more openings or windows", unambiguously requiring openings on the outer surface.
The protrusion allows positioning of photodetectors within or beneath it.	Low	The text describes this as a possibility but does not strictly require internal photodetector positioning.
The protrusion comprises opaque material.	Low	While some sections describe opaque material, it is not strictly required in all protrusion embodiments.
The opaque material of the protrusion extends from its convex surface to its opposite interior surface.	Low	This describes one embodiment but does not necessarily require the full extent of opaque material.
At least a portion of the protrusion comprises opaque material.	Low	While some embodiments specify opaque material, it is not required in all embodiments.
The protrusion comprises opaque material to prevent light piping.	Low	The purpose of opaque material is described but not necessarily required.

Function: “Protrusion”

Structure	Confidence	Analysis
Conforms tissue of the measurement site	High	The text states the protrusion is "shaped to conform the measurement site into a flat or relatively flat surface", directly indicating tissue conformation is a function.
Provides contact surface	High	The text states the protrusion "includes a measurement site contact area that can contact body tissue", unambiguously requiring a tissue contact surface.
Reduces mean optical path length	High	The text states "the protrusion is used to thin out the measurement site" to reduce path length and increase SNR, clearly linking path length reduction to the protrusion.
Increases optical transmittance	High	The text states the protrusion results in "an intensity gain" and "improvements in the visible/near-infrared range" in transmittance, directly attributing increased transmittance to the protrusion.
Provides openings to pass light	High	The text states "openings or windows on an external surface of the protrusion", clearly requiring openings as a function.
Focuses light onto the photodetectors	Medium	Some embodiments describe light focusing properties but other embodiments do not.
Positions photodetectors	Low	Some embodiments describe positioning detectors but it is not required in all cases.
Reduces light piping	Low	Some embodiments describe opaque material to reduce light piping but it is not universally required.
Thermally manages emitters	Low	While some embodiments describe heat sinks and thermal dissipation, these features are not strictly required in all cases.
Electrically shields photodetectors	Low	Conductive shielding is described in some embodiments but not necessarily required.

Based on what we found in the intrinsic evidence, there appears to be various species of protrusions, one of which is a solid, transparent lens. However, our system suggests that a reasonable construction of the term protrusion for claims 8 and 12 requires a different species, one that is a physical component of the recited housing and, moreover, has a convex surface and has an opaque portion. The surface has openings that extend from the surface through the protrusion itself. These openings are arranged such that they can channel light from the openings through the protrusion to the photodiodes. Additionally, our system identified with high confidence that the protrusion has a role in conforming tissue or adjusting tissue thickness to reduce mean path length, a critical function for the accuracy of physiological measurements.

We discovered Patented.ai engine’s programmatic extraction of term glossary and analyses saved tons of time because we needed only to spot check, rather than read through volumes of specifications, office actions, responses, petitions, and other intrinsic evidence, as one would previously need to do. Moreover, the methodical nature of the review and analyses meant that the term glossary was likely to be exhaustive and, hence, it’s unlikely that there’s intrinsic evidence inconsistent with or contradictory to our understanding of the structure and function of the protrusion. In essence, our system gave us a quick and practical way to prove a negative, i.e., that there is nothing contradictory to our constructions in the intrinsic evidence.

One observation to note: having a term glossary made it much easier to spot ambiguous or vague terms, which can be a good predictor of contention in a claim construction, all things being considered.

Product Landscape: Before and After the Priority Date

By understanding the claim scope, Patented.ai’s Product Discovery technology identified related products, e.g., those that incorporate one or more elements of the claims, so that we could understand the landscape in which the ‘648 patent was examined and operated. What we found suggested there was a variety of related products offered before the priority date and, notably, a much more significant set after the priority date.

Shown below is one example of Patented.ai’s lists of related products available before the priority date. As shown, there was a range of devices, from early pulse oximeters to more sophisticated health monitoring systems, each contributing to the evolution of wearable medical technology.

Automated Product Discovery Results

Predates August 3, 2009

Product	Company
Nellcor N-65	Medtronic plc
BCI 3301	ICU Medical, Inc.
Invivo Escort M8	InVivo Therapeutics, Inc.
WristOx 3100	Nonin Medical, Inc.
Philips IntelliVue MP2	Philips North America LLC
GE TuffSat	GE HealthCare Technologies, Inc.
Santamedical Generation 2	Gurin Products LLC
Viiiiva Heart Rate Monitor	4iiii Innovations Inc.

An example summary is provided below. With over 200 products found by our system from nearly a hundred companies across various industries, it is evident that the landscape has experienced substantial growth and diversification since the priority date of the '648 patent.

Automated Product Discovery

Target Patent:	10,945,648
Date of Patent:	March 16, 2021
Assignee	Masimo Corporation
Products Found	264
Companies Found	93
Industries Found	31

Given the wealth of results, we found our system’s ability to analyze the mountain of documents returned was both effective and useful, helping surface more related products while also saving significant time in reviewing false positives.

Related Printed Publications

As part of our endeavor to get a holistic picture of the patent landscape in which the ‘648 patent operates, we searched and analyzed printed publications. As shown below in a summary of one of our searches, there were myriad documents to review, which we did not have to do, thanks to our system. The measures of confidence in the analysis, as well as the accuracy of conclusions, were also helpful.

Automated Prior Art Search

Target Patent	10,945,648
Date of Patent	March 16, 2021
Priority Date	July 29, 2009
Assignee	Masimo Corporation
Patents Found	527

Printed publications we surfaced include early innovations in pulse oximetry and advancements in health monitoring systems contemporaneous to the ‘648 patent (listed below). As with products, our system’s capabilities saved us from sifting through a ton of false positives. It also helped us gain a solid view of the landscape as known to a POSITA at the priority date.

Automated Prior Art Search

Patent	Title	Date
US-6339150-B1	Mutant rex genes encoding transdominant repressors of HIV/HTLV replication	1989-07-07
US-7349726-B2	Pulse oximeter and sensor optimized for low saturation	1994-04-01
US-7320875-B2	Hepatitis C virus E2 binding protein	1995-09-01
US-6583110-B1	Use of agents which bind G proteins for treating septic shock	1997-09-05
US-6314058-B1	Health watch	1997-11-21
US-6472142-B1	Methods and means for inducing apoptosis by interfering with Bip-like proteins	1997-12-03
US-8443761-B2	Veterinary procedure table with scale	1997-12-16
US-7542878-B2	Personal health monitor and a method for health monitoring	1998-03-03
US-7299159-B2	Health monitor system and method for health monitoring	1998-03-03
US-7455963-B2	Antibodies to cyclin E2 protein	1998-06-05
US-6943238-B2	Antibodies to cyclin E2 protein	1998-06-05
US-6266547-B1	Nasopharyngeal airway with reflectance pulse oximeter sensor	1998-09-09
US-6253098-B1	Disposable pulse oximeter assembly and protective cover therefor	1998-09-09
US-6519487-B1	Reusable pulse oximeter probe and disposable bandage apparatus	1998-10-15

Potential Invalidating Evidence?

Our work revealed several combinations of printed publications and products that are of interest because they may prove material to the issues in dispute. Our results for a wearable device that measures oxygen or oxygen saturation levels fall into three categories:

There were numerous devices found to have a controller to process signals and calculate oxygen and oxygen saturation levels or a strap that appropriately positions the devices for proper measurement, though for the sake of brevity, we won’t discuss them here.

Included in the first category is a thesis entitled, “Analysis and Validation of an Artifact Resistant Design for Oxygen Saturation Measurement Using Photo Pletyhsmographic Ring Sensors,” by Phillip Andrew Shaltis, published by MIT in 2004. See Figure 8-3, which shows an opaque surface of the device with an opening aligned over a photodiode.

Reference

Shaltis (2004)

overlap with '648 · Claims 8 & 12

Claim 8: Protrusion with convex surface and aligned openings

Claim 12: Oxygen saturation measurement

Reasoning

Demonstrates a device design with an opaque surface and photodiode alignment, pertinent to '648's claims of protrusion design and noise reduction due to opaque material. Aligns with the concept of light emission and reception in '648.

See also FIGs. 4-4 and 4-7 of Shaltis (reproduced below), both of which show the ring device having multiple photodiodes in different configurations. Moreover, the latter figure shows a “pusher” that creates a protrusion that adjusts tissue thickness to improve accuracy of measurement. Also interesting is Section 8.2.1, which discusses the advantages of its Ring Sensor in comparison to the Nellcor Fingertip Pulse Oximeter.

Reference

Shaltis (2004)

overlap with '648 · Claims 8 & 12

Claim 8: Multiple photodiodes; Protrusion design

Claim 12: Oxygen saturation measurement

Reasoning

Shows configurations of multiple photodiodes and a protrusion for tissue thickness adjustment.

Reference

Shaltis (2004)

overlap with '648 · Claims 8 & 12

Claim 8: Multiple photodiodes; Protrusion design

Claim 12: Oxygen saturation measurement

Reasoning

Shows configurations of multiple photodiodes and a protrusion for tissue thickness adjustment.

Also included in the first category are Masimo’s own LNOP reusable fingertip sensor, which was for sale in 2000 (see photo below) and US Patent No. 7,764,982 B2 issued to Dalke et al., which published on September 21, 2006.

Reference

Masimo's LNOP Sensor (2000)

overlap with '648 · Claims 8 & 12

Claim 8: General relevance to device design and functionality

Claim 12: Oxygen saturation measurement

Reasoning

Provides early implementation of similar technology.

FIG. 32C of Dalke seemed interesting and is reproduced below. (See, e.g., items 3200 and 3210.)

Reference

US Patent No. 7,764,982 B2 (Dalke et al.)

overlap with '648 · Claims 8 & 12

Claim 8: Protrusion and photodiode arrangement

Claim 12: Oxygen saturation measurement

Reasoning

Illustrates a device with features regarding the arrangement of LEDs and photodiodes.

Also intriguing is US Patent No. 7,355,688 B2 issued to Mao et. al., which describes another product, specifically the ODISsey™ Tissue Oximeter. Mao describes a sensor unit that houses two light sources and four detectors, an example of which is depicted in FIG. 9, reproduced below. The sensor unit has openings aligned with the photodiodes to pass light emitted from the light sources and openings to receive light reflected from tissue. Notably, tissue thickness adjustment and conformation appear to be achieved with a cushioning layer positioned on the far side of the sensor unit with respect to the tissue sampling area. Mao explains that LEDs and photodetectors may be located inside its sensor unit:

In another implementation, the radiation sources, photodetectors, or both may be located at the sensor unit. For example, the source structures may include light-emitting diodes (LEDs) and the detector structures may include photodiodes.
Mao, at col. 18, lines 6 - 9. 

Reference

US Patent No. 7,355,688 B2 (Mao et al.)

overlap with '648 · Claims 8 & 12

Claim 8: Photodiode arrangement and sensor unit design

Claim 12: Oxygen saturation measurement

Reasoning

Shows a sensor unit design with aligned openings and photodiodes.

Here’s a photo of the sensor unit of the ODISsey™ Tissue Oximeter.

Reference

US Patent No. 7,355,688 B2 (Mao et al.)

overlap with '648 · Claims 8 & 12

Claim 8: Photodiode arrangement and sensor unit design

Claim 12: Oxygen saturation measurement

Reasoning

Shows a sensor unit design with aligned openings and photodiodes.

For the second category, we found to be interesting US Patent No. 6,760,609 B2 issued to Jacques. FIG. 1A of Jacques shows an oximeter 100 that is worn on a user’s fingertip. (See below.)

Reference

US Patent No. 6,760,609 B2 (Jacques)

overlap with '648 · Claims 8 & 12

Claim 8: LED and photodiode configuration

Claim 12: Oxygen saturation measurement

Reasoning

Details an oximeter with multiple LEDs and photodiodes configurations.

Moreover, Jacques discloses multiple LEDs at multiple places, for example, see FIG. 11 below and its corresponding text, which discloses light sources 310A, 310B, and 310C, which can be LEDs:

In another improved embodiment of the invention, illustrated in FIG. 11, additional measurements are preferably obtained using more than two light sources 310A, 310B, 310C of different wavelengths in the visible to near-infrared wavelength range. Shorter wavelengths probe the vessel wall 260 and longer wavelengths probe into the blood-filled vessel lumen 270.
Jacques, at col. 25, lines 52 - 58.

Reference

US Patent No. 6,760,609 B2 (Jacques)

overlap with '648 · Claims 8 & 12

Claim 8: Multiple LEDs of different wavelengths

Claim 12: Oxygen saturation measurement

Reasoning

Illustrates an oximeter using multiple LEDs of different wavelengths.

FIGs. 10A and 10B and their corresponding text further disclose several implementations that use 4+ photodiodes that capture light from light sources at different measurement sites:

FIGS. 10A and 10B, according to these improved embodiments, for example, a number of light sources 210 and/or a number of detectors 220 in an oximetry device 200A, 200B are used to provide a range of source-detector separations. The light sources 210 (and/or the detectors 220) can be arranged, for instance, in an array 208. The circuitry that actuates the light sources 210 and circuitry that processes the output signals from the detectors 220, are not shown in FIGS. 10A and 10B, but can be routinely designed by those skilled in the art.
Jacques, at col. 24, line 62 - col. 25, line 6.

Reference

US Patent No. 6,760,609 B2 (Jacques)

overlap with '648 · Claims 8 & 12

Claim 8: Multiple photodiodes

Claim 12: Oxygen saturation measurement

Reasoning

Exhibits implementations using multiple photodiodes and their role in accurate physiological parameter measurement.

And Jaques explains that multiple measurement sites are advantageous:

Furthermore, the optical properties of the surrounding tissue may vary. The thickness of the vessel wall may also vary from patient to patient and the size of the inner lumen may vary. Given these additional variables, it is desirable for an embodiment of the invention to conduct more measurements in order to specify the oxygen saturation. To accomplish this, it is desirable to expand the number of measurements that are included in the oximetry method to provide accurate measurements in more complex tissues.
Jacques, at col. 24, lines 52 - 54. (Emphasis added.)

For the third category, we thought OXIMET’s device was interesting as it shows, in the figure below, glass windows that separates light sources and sensors from samples.

Reference

OXIMET Device

overlap with '648 · Claims 8 & 12

Claim 8: Window design in photodiode protection

Claim 12: Oxygen saturation measurement

Reasoning

Features glass windows separating light sources from sensors, relevant to '648's claims on window design for photodiode protection and accurate light transmission.

Also included in the third category is the Shaltis’ Ring Sensor (referenced above), which has a transparent window that covers the opening to protect the photodiode from contact with tissue. (See Figure 4-4 of Shaltis’ dissertation below.)

Reference

Shaltis (2004)

overlap with '648 · Claims 8 & 12

Claim 8: Transparent window design

Claim 12: Oxygen saturation measurement

Reasoning

Illustrates a transparent window covering the photodiode, relevant to '648's claims on window design to prevent direct tissue contact and reduce noise interference in measurements.

We hope you find these findings as interesting and compelling as we did. While we offer no legal opinions, we believe our technology has revealed insights, many of which may be material to the issues in the dispute and that traditional methods might have missed or, at a minimum, taken significantly longer to uncover.

The Power and Potential of Patented.ai’s Technology

What have we learned from testing Patented.ai’s engine in this real-life dispute? Most importantly, our technology accelerates the iterative process of patent analysis, leading to quicker clarity and a solid grasp of the patent landscape. It also underscored the importance of pressure testing the system with complex cases like Masimo vs Apple, which exceed the average complexity by several standard deviations, to identify any weaknesses and shortcomings. We have witnessed first-hand the effectiveness of what we’ve built to date and are committed to applying our learnings to enhance our AI engine and its capabilities.

Programmatic detection and review have significantly reduced the manual labor customarily required for rigorous patent analyses. The AI engine’s ability to extract exhaustive glossaries of claim terms of interest spared us from reading through mountains of documents. But it’s not just about efficiency – our system searches beyond the customary corpora and found things of interest we believe would have remained undiscovered, regardless of resources.

Most who do patent work, especially those accountable for filing post grant challenges, are likely aware that it is iterative work. For example, one searches and analyzes prior art and repeats until there’s some kind of convergence, usually when the references look similar or come from similar sources, in other words, when sound grounds for invalidity become evident. By leveraging our system’s ability to search and analyze programmatically, we were able to iterate much faster than is customary. Thus, Patented.ai can rapidly achieve clarity and attain a solid grasp of the patent landscape under which patent claims of interest operate. With our technology, what would otherwise take weeks and months, we can accomplish in a handful of days.

As to the row between Apple and Masimo, it’s as complicated a case as we’ve seen. The ‘648 patent’s extensive prosecution history and numerous priority claims rival those of the Lemelson portfolio. Regardless, there appears to be products and printed publications not yet considered by both Masimo and Apple, some of which may benefit either party and potentially alter the course of litigation.