A constrained disorder principle-based second-generation artificial intelligence digital medical cannabis system: A real-world data analysis

Abstract

Introduction:

Adhering to treatment plans can be challenging for medical cannabis patients. According to the constrained-disorder principle (CDP), biological systems are defined by their degree of variability. CDP-based second-generation artificial intelligence (AI) systems use personalized variability signatures to improve chronic medication response.

Aim:

We retrospectively analyzed real-world data regarding chronic pain patients using the second generation of artificial intelligence systems to improve adherence to medical cannabis and increase its effectiveness.

Design and methods:

A retrospective analysis of real-world data of 27 patients using prescribed medical cannabis for chronic pain was conducted. Patients received treatment according to a regimen provided by the CDP-based second-generation AI Altus Care™ app that managed the product’s dosage and administration times. The app offers a therapeutic regimen by varying dosages and administration times within predefined ranges. We included 16 patients who participated for more than a week. We assessed adherence to therapy and clinical response in real life based on pain scale measurements.

Results:

The patients were followed up for 64 days (30–189). Second-generation, AI-based, personalized regimens had a high engagement rate and adherence. 50% of patients showed a high compliance rate. Chronic pain improved in patients who reported their pain score.

Summary:

This preliminary real-world data analysis suggests that an algorithm-based approach using a second-generation AI system may enhance the adherence to and clinical effectiveness of medical cannabis. These findings require confirmation through prospective controlled studies.

Relevance for public health

The increasing use of medical cannabis encounters various challenges, including the absence of standardized treatment regimens, low adherence rates, and significant variability in responses both between individuals and within the same individual. We conducted a retrospective analysis of real-world data concerning chronic pain patients to assess the effectiveness of second-generation artificial intelligence systems in improving adherence to medical cannabis usage. Our findings indicate that personalized regimens based on second-generation AI achieved a high engagement rate and adherence. Fifty percent of patients demonstrated strong compliance. Notably, chronic pain improved in patients who reported their pain scores. This real-world data suggests that an algorithm-based approach utilizing a second-generation AI system can enhance both adherence to medical cannabis and its clinical effectiveness.

Introduction

Various medical cannabis formulations are used for numerous indications, including chronic pain, inflammation, autism, sleep disturbances, and many more.^1–5 Chronic use of cannabis-based products is plagued by low adherence and decreased effectiveness. ⁶ Formulations, therapeutic regimens, and inter and intra-subject variability in responses make it difficult to standardize therapy and maximize these products’ therapeutic potential. ⁷ Cannabis is one of the most widely used recreational psychoactive substances in the world. In 2018, it was estimated that there were 192 million cannabis users, which represents 3.9% of the global population aged 15–64 years. Despite ongoing efforts to regulate cannabis worldwide, significant differences in country policies and usage patterns may unpredictably and disproportionately increase the prevalence of cannabis use disorders. ⁸

Patients and doctors demonstrate low engagement with digital applications aimed at improving adherence. A recent review of studies that examined medical apps for enhancing patients’ responses showed that among 17 studies that demonstrated some level of effectiveness, eight showcased statistically significant improvements. ⁹ Only a single study showed improved objective measures beyond relying on patient-reported questionnaires. ⁹

Medication management apps improve patient education by allowing healthcare providers to educate patients on the importance of medication adherence and effective medication management. ¹⁰

The Constrained Disorder Principle (CDP) describes biological systems based on their inherent variability.^11,12 All systems are characterized by a dynamic level of variability, which facilitates better adaptation to both internal and external disturbances. According to the CDP, malfunctions in systems occur due to either insufficient or excessive variability.^13–15

The partial or complete loss of response significantly contributes to low adherence to chronic medications. ¹³ CDP-based second-generation artificial intelligence (AI) systems implement variability into therapeutic regimens within a predefined therapeutic range to improve the effectiveness of chronic medications and overcome the loss of response.^10,16,17

The term “digital medical cannabis” refers to any cannabis formulation whose administration is controlled by second-generation artificial intelligence systems. ⁷ Patients with chronic diseases benefit from the system by improving clinically meaningful outcomes, such as pro-BNP levels, 6-min walk tests, and Kansas City cardiomyopathy scores in congestive heart failure.^7,16,18–39

The paper summarizes retrospective preliminary proof of concept real-world data of chronic pain patients receiving digital medical cannabis. In real-world settings, we examined the effect of an algorithm-based regimen on cannabis adherence and effectiveness. To assess compliance, we quantified the frequency of patient logins to the app to monitor their treatment plan. Effectiveness was evaluated through the pain rating recorded within the app.

Design and methods

Retrospective analysis of real-world data: Data from patients receiving medical cannabis for chronic pain was analyzed retrospectively and non-interventionally. The study analyzed data of individuals who received cannabis from Hadassah Ein Kerem Hospital in Jerusalem throughout 2022. Their physicians offered subjects to use the app for their cannabis usage.

Data collection: Patients’ data is anonymously collected in the cloud and analyzed retrospectively. Several self-entered outcome measures were followed. The pain scale ranged from one to ten, where a lower score indicated no pain while a higher score indicated the most intense pain. The percentage of app logins to the total planned logins was measured; a patient receiving treatment three times a day was calculated as the number of logins divided by three times the number of follow-up days. Improvement in pain was defined as any decrease in the pain scale throughout the follow-up period. Each patient served as their baseline, allowing for a comparison of their progress from the beginning of the trial. Moreover, good compliance was defined as logging into the app for more than 40% of the planned logins based on previous studies showing low adherence to these regimens. ⁴⁰

Second-generation AI system: Altus Care™ is a cellular phone-based product of Area9 Innovation Apps, part of Area9 Group, that allows easy digitization of treatment plans or research protocols and remote implementation of these plans. The Altus Care™ platform has been combined with treatment algorithms that use second-generation AI to provide random alterations in the dosing and times of administration of medications within a physician’s predefined range and serves as a reminder for patients to take their medications (Oberon Sciences, Israel). Israel’s Ministry of Health has approved its use.

Treatment regimens provided by the app: Physicians prepared a predefined range of minimal and maximal daily dosages and timing frames for each patient. In the app, physicians provide a therapeutic regimen, including the cannabis formulation, the dose range, and the timing of administration. The daily dose and frequency were set to not exceed the patients’ dose before using the app. The Altus Care™ app was downloaded on the subjects’ cellular phones. Reminders are sent to patients to take the designated formulation within the predetermined range. Patients can use the app to answer a daily pain scale questionnaire that their physician can view.

Results

Subjects

The universal sampling data of 27 patients who used medical cannabis according to a regimen provided by the app was analyzed. The analysis did not include patients who used the product for less than a week.

The demographics of the patients in the study are presented in Table 1. Sixteen patients who used the app for more than a week were analyzed, including six males and 10 females, with a mean age of 47. The diagnosis for which subjects received the cannabis included four fibromyalgia, three Back pain, two neuropathy, two Crohn’s disease, three joint pain due to different autoimmune diseases, one chronic pain syndrome after physical trauma, and one chronic abdominal pain without a diagnosis. Data were available for a median follow-up of 64 days (range 30–189). In half of the patients, cannabis oil drops were taken, while in the other half, cannabis was smoked. Eight patients took the medication twice daily, five patients once daily, and three thrice daily.

Table 1.

Demographics of patients.

Gender (male/female) No.		6/10
Age mean (range)		47 (19–73)
Follow-up days, median (range)		50 (30–189)
Administration routes (drops/smoke) No.		8/8
Diagnosis (number, %)	Fibromyalgia	4 (25%)
	Back pain	3 (18.75%)
	Crohn’s	2 (12.5%)
	Joint pain	3 (18.75%)
	Other	2 (12.5%)

The therapeutic effect and adherence to medical cannabis when using the app

A high engagement rate was associated with the use of the app. Figure 1 shows the overall engagement of chronic pain patients with the app. Patients were highly compliant when using the app each day at least once. The app was used at least once daily by 50% (n = 8) of patients. Four patients (25%) had a compliance rate of over 60%, one patient (6.25%) had a compliance rate of 51%, and three patients (18.75%) had a compliance rate between 40% and 50%. The remaining patients had a compliance rate between 10% and 40%. The mean compliance was 40% (range 13%-83%).

Figure 1.

The use of digital medical cannabis leads to better therapy adherence. An analysis of the data from patients using digital medical cannabis shows the rate of treatment adherence.

Chronic pain patients’ clinical outcomes were improved by introducing variability in dosing and administration times in some subjects. Examples of patients’ daily pain scores using the digital medical cannabis app can be seen in Figure 2. Although this is not a controlled study, the app-based regimen data suggested that some patients reported improved pain scores, which further motivated adherence to the treatment.

Figure 2.

Examples of how digital medical cannabis can help alleviate chronic pain include patient-reported pain scores collected retrospectively. Each panel in the data represents an individual patient. (a) A favorable and sustained clinical effect was observed within a few weeks of use. (b and c) There is variability in the clinical effects experienced by different patients.

Discussion

Despite the increasing use of medical cannabis for multiple purposes, low adherence and loss of response remain significant challenges for its chronic use. ⁷ The preliminary data presented here show that digital medical cannabis, an app-regulated personalized therapeutic regimen, may overcome some challenges associated with medical cannabis. The second-generation AI system-based app regulates the use of all types of cannabis products. The system controls doses and times of administration utilizing a variability-based algorithm. While this study is not a controlled trial, the preliminary data may suggest that this system improves adherence and clinical outcomes. Combining drugs with devices may enhance adherence and outcomes, motivating patients to follow their treatment regimens.

With the legalization and increased use of medical cannabis, various products became available. The pharmacokinetics of cannabis formulations vary markedly between and within subjects. ⁴¹ Although cannabis is a highly personalized medicine that requires titration, there is no method to personalize the therapy. Most therapeutic regimens are selected based on trial and error rather than validated data. There is a great deal of difficulty for physicians and patients in maximizing the benefits of these products.

Using medical cannabis for an extended period is difficult due to low adherence. In a population-based study of 5452 new users, the median use duration was 31 days, and only 18% used cannabis for 1 year. ⁴² A survey of 4000 subjects consuming cannabis at least once in the past year classified the subjects into four groups based on the number of usage days per year as follows: infrequent users (<11 days), occasional users (11–50 days), regular users (51–250 days), and intensive users (>250 days). ⁴³ Additionally, clinicians fail to follow prescription guidelines, complicating standardizing therapeutic regimens. ⁴⁴

A common problem with chronic drugs is a loss of response to them. ¹⁶ As a result of repeated exposures to cannabis, tolerance and effectiveness are reduced.^45–47 As a result of chronic use, tolerance develops, leading to a partial or complete loss of the clinical effect.^47–49

A randomized, placebo-controlled crossover study assessed how cannabis affects the brain in occasional and chronic users and found a pharmacodynamic mechanism for tolerance development.^45,47 Occasional users displayed significant neurometabolic changes in reward circuitry, decreased functional connectivity, increased striatal glutamate concentrations, and increased subjective high. Chronic cannabis users did not appear to experience similar changes, suggesting that the reward circuitry is less responsive to cannabis intoxication than chronic cannabis users.^45,47

It has been proposed that the biological response to medical cannabis takes on a U-shape. ⁵⁰ Titration should begin low and be gradual to achieve better results. When a dose is raised beyond a certain level, there is a general tendency to induce tolerance, a loss of effect, and more adverse effects.^49,51,52

The first generation of AI primarily focuses on clinical decision-making through big data analysis. As a result, their real-world use is limited since most algorithms do not necessarily lead to better outcomes for patients.^53–56 These algorithms may be affected by biases resulting from using big data, which may negatively influence the overall outcome of this analysis. The engagement rate of patients and physicians in medication reminder systems is low.^57–60 Using mobile phones to remind patients to take their medications is insufficient to improve adherence to medication.^61–63

Second-generation AI systems based on CDP focus on improving patient outcomes through subject-specific AI. ¹¹ Their development evolved from biological systems relying on regulated noise levels to function.^{13,32,64–72} By overcoming the compensatory mechanisms associated with tolerance and disease progression, focusing on patients’ clinical benefits ensures enhanced adherence and sustainable response to chronic drugs.^16,18–20 An algorithm based on individualized variability signatures is introduced into these systems to enhance the efficiency of chronic drugs individually. The response to chronic therapies is improved by intermittent dose adjustments and drug holidays.^{7,10,11,13,16–39,73–89}

 Digital medical cannabis is provided to subjects through a user-friendly app downloaded to their cell phones. ⁷ It allows the individual to follow a personalized therapeutic plan that enhances adherence and may improve responses to cannabis products by reducing tolerance and improving clinically meaningful outcomes, thereby ensuring greater engagement from both patients and physicians.^7,16,18–20

Three steps are involved in implementing digital medical cannabis. As a first level, an open-loop system was utilized on the subjects in the present study to provide random dosing and administration times within a predefined therapeutic range. ¹⁹ Secondly, a closed loop is implemented, where dosages and administration times are personalized based on patient and provider feedback. In the third level, variability signatures, such as heart rate variability and cytokine secretion variability, are quantified and integrated into the algorithm.^{10,11,16–20,73,84}

One limitation of the current data analysis is that it is based on a small, uncontrolled group of participants and has a relatively short follow-up period. A controlled study, on the other hand, would include a control group that uses an app to remind them to take medical cannabis according to a regular schedule, and it would involve a longer follow-up period.

In summary, this preliminary analysis of real-world data suggests that using a second-generation personalized AI algorithm to randomize cannabis regimens could improve clinical responses and treatment adherence. Digital medical cannabis has the potential to enhance clinical effectiveness and reduce tolerance to cannabis products, which may lead to lower dosages and fewer side effects. However, these findings need to be validated through extensive, prospective controlled studies.