FROM THE PROMISE of tailored drug delivery to augmented reality and continuous health monitoring, there have been numerous avenues of research investigating the high-tech contact lenses of the future. Indeed, for the past several years, Contact Lens Spectrum has regularly published columns on contact lenses of the future and what they may entail. This article will discuss some of the recent developments in powering contact lens electronics, new methods of contact lens manufacturing, and how artificial intelligence (AI) is being leveraged to help design contact lenses.
Electronics and Power Delivery
Several proof-of-concept contact lens sensors have entered clinical trials or been commercialized.1 The most prominent example uses an internal strain gauge to measure changes in the curvature of the cornea, which correlates to intraocular pressure (IOP).2 The information is transmitted wirelessly to a recording device, providing an opportunity for a large amount of data to be collected.2
Despite promising theoretical results, the challenges in the contact lens sensor space are numerous. There are engineering challenges—from the ability of the device to measure the target variable to the miniaturization and packaging in a contact lens—to ensuring biocompatibility and acceptability to the patient. Considerations also have to be made regarding cost and large-scale manufacturing if these devices are to be ultimately commercially viable.
Some of the recent developments in the contact lens sensor space have focused on the materials used for the circuits embedded into the lenses. An example of a class of materials being explored are the “MXenes,” which are transition metal carbides, nitrides, or carbonitrides.3,4
These materials are of interest for contact lenses because, in addition to their electrical properties, they are also hydrophilic, have tunable optical properties, and are mechanically flexible.4 The materials have previously been incorporated into intraocular lenses, where their electrical conduction properties are leveraged to induce a change in the focus of the lens when a current is applied.4
These MXenes may also absorb different wavelengths of the electromagnetic spectrum—a phenomenon that has generated interest in incorporating these materials to help block UV rays. This ability could be used to protect against electromagnetic (EM) interference generated by high-tech, smart contact lens systems that use radiofrequency protocols for power or communication, reducing ocular exposure.3
For example, in a series of experiments in coating commercially available lenses with an MXene based on titanium, researchers were able to maintain at least 80% transmission of light in the visible spectrum while demonstrating 47% to 93% blockage of EM wave transmissions in the 5.8 GHz range, which is commonly used for WiFi, depending on the concentration of the MXene coating used.3 They also tested the lens in a porcine ex-vivo model to absorb microwaves, demonstrating negligible increases in the temperature of the ocular surface when exposed to these rays, with the contact lens itself heating up instead as it absorbed the microwave energy.3 The research also demonstrated some evidence that the coating may also help prevent dehydration of the lenses when exposed to air.3
Power delivery is also crucial for any electronic application and is more challenging considering the size of a contact lens and the sensitivity of the eye.5 Conventional batteries cannot be used in this context due to concerns regarding toxicity in their components, heat generation during use, and the rigidity of the materials needed.5
Powering of contact lens devices has thus relied on either direct connection to an external power source via a conductive wire or wireless power transmission, with each presenting its own challenges. A direct connection with an external power source has the advantage of a stable flow of energy compared to other methods and can also be coupled to transmit data from sensors if necessary. Less of the contact lens needs to be devoted to the power supply with a direct connection, allowing for more space to incorporate other components. The downside of a direct connection is that the lens may be uncomfortable, which impacts its long-term wear potential.5
In contrast, wireless power delivery does not require a direct connection to the power source but does require a built-in antennae to receive the power and a nearby transmitter to produce power. The frequency of power transmission also impacts performance: Higher frequencies often require smaller antennae but generate more heat, while lower frequencies are more efficient in generating power but require larger antennae.5
A recent report of a hybrid power system for a contact lens sensor application leverages both the solar capture of energy as well as a metal-air harvester based on magnesium and oxygen, and it is powered by the action of the blink.6 This technology relies on the tears as an electrolyte and a source of oxygen to connect the cathode and the anode of the circuit, which occurs during the blink.6 The rest of the time, the electrolyte does not connect the 2 terminals, which has the advantage of significantly increasing the lifetime of the battery as it does not foul as easily, resulting in an estimated lifespan on the eye of approximately 34 days.6 The constant replenishment of the tears on the ocular surface provides the dissolved oxygen needed to power the electrical circuit.
The researchers combined this blink power generation with a silicon solar cell bank that can capture light in both outdoor and indoor conditions.6 Importantly, both the blinking and solar methods of power generation do not require external power accessories beyond what is already on the eye; they also only deliver pulsed energy, depending on the lighting conditions and blink rate.6 To ensure that a stable voltage is delivered to the onboard electronics when needed, an integrated power management and storage solution was developed using capacitors.6 Ultimately, the system is reported to be able to deliver approximately 149 mW at 3.3 volts of direct current output, sufficient to power numerous sensors of the size needed on contact lenses.6
High-Tech Manufacturing
Additive manufacturing methods, also known as 3D printing, have been proposed as a method to manufacture individual or personalized contact lenses. In contrast to traditional spin-casted, molded, or lathed lenses, 3D printing may have advantages in customizability, being able to create complex geometries that would be difficult, if not impossible, through traditional methods.7 These could also be manufactured in a repeatable fashion, with less required post-processing.7 Multiple objects or duplicates can also be manufactured at the same time.7 The challenge to 3D printing centers on the selection of the appropriate material to ensure that the ensuing lens has the appropriate optical and mechanical properties.7
Researchers have reported success using a combination of digital light printing, which utilizes light to cure the material, and a clear, commercially available resin based on methacrylate and diphenyl (2,4,6-trimethylbenzoyl) phosphine oxide.7 Contact lenses were not only manufactured with appropriate optical transparency and mechanical properties—over 90% transmission in the visible spectrum and adequate modulus for handling—but also with complex geometries within the lenses, such as microchannels.7
The researchers have also incorporated dyes within the 3D printing process for tinted lenses that could to help color discrimination in those with color deficiency.8 By incorporating Atto 565 and Atto 488 dyes into the 3D-printing process, undesirable light wavelengths in the 550 nm to 580 nm and 480 nm to 500 nm were filtered out, which could allow for better color discrimination in those who have red-green color vision deficiency.8
AI and Contact Lenses
With the explosion of the application of AI in multiple areas of health care, the technology has also been studied for contact lens design, particularly for more difficult cases such as the management of keratoconus or orthokeratology (OK).9-11 There have been several studies that have incorporated AI into the retrospective examination of these fits, to compare how well AI models are able to predict key contact lens parameters.
In a retrospective analysis of 197 keratoconic eyes fit with rigid lenses, 3 types of AI models—including convolution neural networks (CNN), multilayer perception, and standard machine learning—were compared to standard methods such as mean K or the fitting guide from the manufacturer.9 The AI models were also fed key indices from topographical maps or, in the case of the CNN model, the topographical maps themselves (sagittal, tangential, and corneal thickness maps).9
The CNN model that had access to the topographical maps was deemed to be the most accurate and came closest to matching the parameters of the lenses that were ultimately fit to the eyes. The majority of the AI models were also more likely to match the final order of the back-optic zone radii compared to initial lenses based on mean K.9
AI was also used to analyze 797 eyes that were later fitted with OK lenses that had adequate results per topography.10 The AI models were fed spherical refraction, keratometry, eccentricity, corneal astigmatism, horizontal visible iris diameter (HVID), inferior-superior index, surface asymmetry index, surface regularity index, and 8 mm chordal height difference. They also compared the suggested parameters for the OK designs to what was suggested by the manufacturer’s fitting method. A strong correlation was found between the AI parameters and the final ordered lens parameters. The AI was also trained to automatically detect the edge of the cornea and determine the HVID based on the topographical data inputted.10
AI models are also being trained to predict optimum OK parameters for myopia control effects.11 In another retrospective study, more than 3,500 OK lens fits for myopia control were followed between 9 and 13 months. By measuring their axial length change as well as their post-OK topographies, the researchers codified lens centration and whether the child was a fast myopia progressor (more than or equal to 0.3 mm per year).11
Eighty percent of the dataset was used to train different AI models to perform a 2-step process. The first was a “recall” step, in which the AI was used to systematically generate multiple potential lens designs based on topography image input. The second “ranking” step evaluated the potential lens designs for 2 categorical predictions: whether the lens would have acceptable centration and whether the lens design was predicted to cause an axial length growth rate of less than 0.3 mm per year. This was then compared to the actual results in the final 20% of the dataset, which was used as a test data set, to see the accuracy of these 2 predictions from the model.11
The models were approximately 95% accurate in predicting whether or not the lens design would have acceptable centration, and approximately 88% accurate in predicting whether or not the eye would have more or less than 0.3 mm/year axial length growth rate.11
It should be noted that these studies on AI and lens design were constrained to retrospective cases that were known to have good lens fit and, where appropriate, good myopia control. The researchers in these studies selected keratoconus cases that were successful with their fits or, in the cases of OK, the cases that had intact red rings on topography, suggesting that successful centration and OK corneal reshaping for those eyes were possible. While this is likely necessary to test the ability of the AI models to design lenses for the optimum cases, future studies evaluating their potential for eyes with less than ideal fits or eyes that were unsuccessfully fit with conventional methods will be of interest.
Summary and Conclusions
The promise of high-tech lenses is vast. Their proponents tout the advantages of continuous IOP monitoring, AI integration, wireless transmission, and localized electrical generation. There are also arguments regarding the integration of AI with lens sensors to create useful datasets and clinical interventions.12 This can provide a tailored bridge to the field of theranostics, for which a sensor output can be directly linked to the tailored or triggered release of therapeutics on demand.12
The challenges to these systems include the precise engineering needed to detect the markers of interest in a reliable and accurate fashion, the longevity of the devices, their biocompatibility, scalability, manufacturing, powering, and transmission of data. The field is still nascent in its clinical application and, understandably, lacking in clinical evidence that their use improves clinical outcomes.
On a more fundamental level, the appropriate application of contact lens-based sensors is not only an engineering problem but also a biological one. Part of the challenge for anyone working in this field is the biological relevance of what is being detected, whether it is clinically meaningful and whether the information it provides can be acted upon.
The developmental pathway of contact lens glucose sensors is illustrative. Vast resources have gone into different attempts to reliably and accurately detect glucose in tears, but their ultimate clinical utility is hampered not by the ability to detect those levels reliably, but rather the variable lag time between changes in blood glucose and tear glucose levels, which can vary from 10 to 30 minutes.13 This presents a significant biological challenge to the application of the data received from contact lens glucose sensors if the information is being relied upon to deliver treatments such as insulin. Some argue that there is utility in the personalized nature of these sensors, as they can be used to determine individual lag times between blood and tear glucose levels, which can then be used to make decisions.13
This is one of many challenges, as well as potential solutions, which are proposed by these high-tech contact lenses. Future work will be of interest to see how these technologies are developed, adopted, and implemented over time.
Acknowledgements: The author would like to thank Ms. Lily Ho, UNSW Sydney, for her kind review and suggestions for this article.
References
1. Dunbar GE, Shen BY, Aref AA. The Sensimed Triggerfish contact lens sensor: efficacy, safety, and patient perspectives. Clin Ophthalmol. 2017;11:875-882. doi: 10.2147/OPTH.S109708
2. Shaker LM, Al-Amiery A, Takriff MS, Wan Isahak WNR, Mahdi AS, Al-Azzawi WK. The future of vision: a review of electronic contact lenses technology. ACS Photonics. 2023;10(6):1671-1686.https://doi.org/10.1021/acsphotonics.3c00523
3. Hu L, Azhari S, Zhang H, et al. MXene-integrated contact lens: a breakthrough in wearable eye protection and healthcare. Small Sci. 2025;5(7): 2400628. doi: 10.1002/smsc.202400628
4. Khosravi A, Zarepour A, Zarrabi A, Iravani S. MXene-based wearable contact lenses: integrating smart technology into vision care. Nanomicro Lett. 2025;18(1):20. doi: 10.1007/s40820-025-01863-5
5. Mirzajani H, Mirlou F, Istif E, Singh R, Beker L. Powering smart contact lenses for continuous health monitoring: Recent advancements and future challenges. Biosens Bioelectron. 2022;197:113761. doi: 10.1016/j.bios.2021.113761
6. Pourshaban E, Karkhanis MU, Deshpande A, et al. Power scavenging microsystem for smart contact lenses. Small. 2024;20(32):e2401068. doi: 10.1002/smll.202401068
7. Alam F, Elsherif M, AlQattan B, et al. 3D printed contact lenses. ACS Biomater Sci Eng. 2021;7(2):794-803. doi: 10.1021/acsbiomaterials.0c01470
8. Alam F, Salih AE, Elsherif M, Yetisen AK, Butt H. 3D printed contact lenses for the management of color blindness. Additive Manufacturing. 2022;49: 102464. doi: 10.1016/j.addma.2021.102464
9. Abadou J, Dahan S, Knoeri J, Leveziel L, Bouheraoua N, Borderie VM. Artificial intelligence versus conventional methods for RGP lens fitting in keratoconus. Cont Lens Anterior Eye. 2025;48(1): 102321. doi: 10.1016/j.clae.2024.102321
10. Zhou X, Zeng L, Shen Y, et al. Artificial intelligence-assisted fitting method using corneal topography outcomes enhances success rate in orthokeratology lens fitting. Cont Lens Anterior Eye. 2025;48(3):102350. doi: 10.1016/j.clae.2024.102350
11. Zhang Z, Zhou X, Li C, Xiao Y, Chen Z. A multimodal multitask artificial intelligence model for orthokeratology contact lens fitting: an integrated framework to enhance lens centration and myopia control effect. Ophthalmol Sci. 2026;6(4): 101100. doi: 10.1016/j.xops.2026.101100
12. Boppiniti ST. Novel smart contact lenses with embedded AI for pioneering continuous health tracking and disease detection. IJIERT. 2024;11(9):1-17. doi:10.26662/ijiert.v11i9.pp1-17
13. Park W, Seo H, Kim J, et al. In-depth correlation analysis between tear glucose and blood glucose using a wireless smart contact lens. Nat Commun. 2024;15(1):2828. doi: 10.1038/s41467-024-47123-9