AI in Healthcare: Transforming Patient Outcomes in 2025

The healthcare staffing crisis has become a global challenge, with AI in healthcare examples emerging as vital solutions. NHS trust staff shortages could reach 250,000 full-time positions by 2030. The global healthcare sector faces an even bigger challenge – a projected shortage of 18 million professionals, including 5 million doctors.

These numbers tell a worrying story, but artificial intelligence brings hope through innovative solutions. AI tools make use of information from large datasets to spot patterns better than humans in many healthcare areas. AI healthcare applications improve accuracy, cut costs, and save time while reducing human errors.

Clinical artificial intelligence does more than just fill empty positions in healthcare. Clinicians can discover new treatment approaches and diagnostic methods that help reduce medical uncertainty. On top of that, these technologies handle routine work and free up clinicians to spend quality time with patients, making care more personal in unexpected ways.

The National Academy of Medicine highlights three benefits of AI for healthcare: better results for patients and clinical teams, lower costs, and improved population health. In this piece, we explore how these AI-powered solutions create real improvements in patient outcomes for medical specialties of all types in 2025.

_{Image Source: NextGen Invent}

AI medical systems are revolutionizing medical diagnostics with unprecedented accuracy levels. The FDA has authorized about 950 medical devices that use AI or machine learning as of August 2024, and this is changing how healthcare providers diagnose diseases.

Medical imaging AI brings remarkable improvements to radiology departments. The latest ultra-low-dose CT (ULDCT) scans powered by AI can spot pneumonia accurately. These scans cut radiation exposure by more than 98% [link_2]. Patients receive only 1.95% of the normal radiation dose with ULDCT (0.12 mSv vs. 6.15 mSv).

AI-enhanced ULDCT delivers perfect accuracy for pneumonia detection in immunocompromised patients, compared to 96-98% with standard ULDCT images. The system spots subtle details that doctors might miss. It achieves 93% accuracy in identifying tree-in-bud patterns versus 78-80% for regular images. The system also detects interlobular septal thickening better, with 78-83% accuracy compared to 61-67%.

Google’s deep learning algorithm shows impressive results in lung cancer detection. Testing on 6,716 National Lung Screening Trial cases showed a 94.4% area under the curve, beating six radiologists. The system reduced false positives by 11% and false negatives by 5%. Sybil, another AI algorithm, predicts future lung cancer risk effectively – with 0.92 area under the curve for one-year predictions and 0.75 for six-year predictions.

Diabetic retinopathy (DR) leads to preventable blindness cases worldwide. The DeepDR system, trained using 466,247 fundus images from 121,342 diabetic patients, spots retinal lesions with high precision.

DeepDR’s performance is remarkable. It achieves area under curve values of 0.901 for microaneurysms, 0.941 for cotton-wool spots, 0.954 for hard exudates, and 0.967 for hemorrhages. The system grades DR severity with AUCs between 0.943 and 0.972 across different stages.

DeepDR Plus takes this further by predicting DR progression within five years. Its concordance indexes range from 0.754–0.846. This helps doctors customize screening intervals, potentially extending average screening time from 12 months to 31.97 months without risking patient safety.

The automated retinal disease assessment (ARDA) algorithm shows AI’s real-world value. In a study of 4,537 patients, it achieved 97.0% sensitivity and 96.4% specificity for severe or proliferative DR. The system didn’t miss any severe/proliferative DR cases, properly flagging them for referral.

AI makes pathology more accurate. A meta-analysis of generative AI models showed 52% overall accuracy, but specific models like GPT-4, GPT-4o, and Claude 3 matched non-expert physicians’ performance. Some models performed as well as experts in specific fields.

AI systems detect breast cancer with 90% sensitivity, while radiologists achieve 78%. Early breast cancer detection shows similar results – AI systems are 91% accurate compared to radiologists’ 74%.

CHIEF (Clinical Histopathology Imaging Evaluation Foundation) represents the latest breakthrough. This versatile AI model, tested on 19,400 whole-slide images from 32 independent datasets, detects cancer with 94% accuracy. It outperforms existing AI approaches across 15 datasets containing 11 cancer types. CHIEF predicts patient survival and treatment responses, marking a big step forward in precision medicine.

These diagnostic applications show how AI in health sets new standards for healthcare accuracy and efficiency. Healthcare providers now have powerful tools to intervene earlier, plan treatments better, and improve patient outcomes.

Treatment planning creates a major bottleneck in healthcare delivery that delays the time between diagnosis and treatment. AI healthcare applications now help solve this challenge by cutting down preparation time for complex procedures.

Radiotherapy planning used to be a time-consuming process. Oncologists had to look at dozens of CT images and draw contour lines around tumors and healthy organs at risk. This contouring process took several hours of a specialist’s time to create a single patient’s treatment plan.

Microsoft Research, Addenbrooke’s Hospital, and the University of Cambridge worked together for eight years to develop InnerEye technology. This open-source AI tool cuts radiotherapy preparation time by up to 90%, so patients wait less time for life-saving treatments.

The results make a real difference in clinical practice. Radiation oncologists who use InnerEye complete image segmentation in about 5 minutes per scan, instead of the 73-87 minutes needed without AI. Here’s the breakdown:

• Head and neck cancer cases: 4.98 minutes with AI versus 73.25 minutes manually
• Prostate cancer cases: 3.40 minutes with AI assistance
• AI takes just 23 seconds to process a full CT scan

Dr. Raj Jena, an oncologist working on the project, calls these results “a game-changer” because faster radiotherapy helps patients survive longer and feel less anxious. The AI models match expert accuracy levels for 13 out of 15 anatomical structures.

Operating rooms bring in substantial money for hospitals, but poor scheduling wastes resources. Research shows that 60% of elective surgeries take less time than scheduled (29 minutes too long), while 37% run longer than planned (30 minutes too short). Operating rooms cost $37-100 per minute ($2,220 per hour), making scheduling mistakes very expensive.

AI healthcare systems analyze patient details, surgeon data, and past results to create better surgical schedules. One machine learning system made predictions 34% more accurate across all surgical departments. Zaribafzadeh’s research showed that AI-assisted scheduling reduced underpredicted cases by 4.3% and increased accurate scheduling within 20% of the actual duration by 3.4%.

AI helps with preoperative planning in several ways:

1. Surgeons made 39.7% fewer adjustments to AI-generated total knee arthroplasty plans compared to standard manufacturer plans, according to Lambrechts.
2. Lopez’s team created a machine learning model that identifies patients who can safely go home the same day after arthroplasty procedures.
3. Li’s research used neural networks to read CT images and create precise specifications for patient-specific knee arthroplasty instruments without adding preparation time.

AI also helps doctors decide if surgery is truly needed. To name just one example, machine learning accurately determines whether patients need hip surgery based on hospital records.

These AI benefits go beyond helping individual patients. Doctors spend less time on repetitive tasks, which helps address staff shortages. Hospitals can treat more patients with their current resources – a vital improvement as healthcare systems face growing patient numbers.

Genomic medicine is at the cutting edge of healthcare breakthroughs, and artificial intelligence tools are taking treatment customization to new heights. AI and genomics working together open up amazing possibilities to tailor medical treatments based on each patient’s genetic makeup.

Transcriptomic profiling has changed the game in precision oncology. AI systems don’t just treat cancer as one disease anymore – they look at gene expression patterns to find different molecular subtypes that need their own treatment approaches.

Medulloblastoma treatment shows this change perfectly. Scientists used AI to analyze hundreds of exomes and found distinct molecular subgroups of the disease. This led to more targeted treatments. Kids with the ‘wingless’ tumor subgroup have benefited the most – they now only need chemotherapy instead of whole-brain radiation that often leads to brain damage and new cancers.

AI models are doing amazing things beyond just classifying cancer. Deep learning combines knowledge from scientific papers with sequencing results to:

• Propose 3D protein configurations
• Identify transcription start sites
• Model regulatory elements
• Predict gene expression from genotype data

These insights are the foundations for connecting genomic variations to how diseases show up, how well treatments work, and what might happen next.

Generative AI models are proving to be great tools in tailored medicine. Studies show that generative adversarial networks (GANs) are the top choice (16 studies), with variational autoencoders (VAEs) coming in second (7 studies). These technologies help deal with data shortages through synthetic data creation, small-dataset learning, and privacy-safe synthesis.

AI’s most influential use in genomic medicine helps predict how patients will respond to specific treatments. The system analyzes gene expression data to figure out which drugs will work best for each patient before treatment starts.

McDonald et al. trained a support vector machine with patients’ gene expression data to predict their response to chemotherapy. The results looked promising across several drugs. Later, Sadanandam et al. created ways to spot patterns in gene sequences that linked to better results from nontraditional treatments.

The predictions are remarkably accurate. A study of gastric cancer cell lines showed a strong positive correlation (R² for CCLE GC: 0.980 and R² for GDSC GC: 0.520) between predicted and actual ln(IC50) values. The ridge prediction model worked best for panobinostat (R²: 0.470 and RMSE: 0.623) compared to other drug models.

Predicting drug responses isn’t easy because it needs different types of data to work together. Scientists came up with clever solutions like DrugS (Drug Response prediction Utilizing Genomic features Screening), which uses gene expression and drug testing data from human-derived cancer cell lines effectively.

These technologies have made a huge difference in clinics. AI models looked at patient data from The Cancer Genome Atlas (TCGA) and spotted drug-sensitive patients successfully. The predicted LN IC50 values were much lower in patients who responded to treatment. Patients with lower LN IC50 values lived longer, too.

AI-powered genomic medicine keeps moving from theory into real-world practice. The National Institutes of Health created an AI tool that looks at individual cells inside tumors to predict if someone’s cancer will respond to specific drugs. This approach gives more detailed data than traditional bulk sequencing, which just averages all cells in a tumor sample. It might explain why some patients stop responding to certain drugs.

_{Image Source: PEPID Pulse}

AI systems have revolutionized remote patient care by taking health monitoring beyond hospital walls. Healthcare providers can now track their patients’ conditions immediately and step in before complications become serious, thanks to connected technologies.

Modern flexible electronic wearables offer benefits that are way beyond the reach of regular monitoring methods. These benefits include cheaper medical costs, quick data access, non-invasive methods, and easy scaling. AI algorithms analyze continuous streams of physical states and biochemical signals from these devices to give customized health insights.

AI helps improve wearable technology accuracy by:

• Finding and fixing errors in collected data
• Recognizing patterns to isolate individual signals
• Collecting synchronized data across body sensor networks
• Using reinforcement learning for energy-efficient routing

Ambient intelligence takes monitoring further with contactless sensors in physical spaces that watch patient conditions without direct contact. These systems can spot falls, track movements, and observe behavior patterns without patient interaction. Acute-care settings now use computer vision and AI cameras to catch medication errors immediately by checking labels, doses, and procedures against patient records.

Regular early warning systems don’t deal very well with clinical deterioration, with positive predictive values at only 5-10%. AI prediction models solve this by analyzing multiple data points to catch subtle warning signs that regular methods miss.

A newer study, published by an academic medical center, showed an AI deterioration model that cut down care escalations by 10.4 percentage points. Patients who received AI alerts were 43% more likely to get the right level of care and had better survival rates during their hospital stay.

The largest longitudinal study proves that when used in real-life clinical settings, AI helped reduce deaths both in hospitals and within 30 days after discharge. While ICU transfers showed a slight decrease (not statistically proven), patients stayed longer in ICU—this suggested earlier treatment for those who truly needed intensive care.

AI-powered remote monitoring helps manage chronic conditions from early detection through ongoing treatment. Smart devices help diabetes patients by monitoring glucose levels and suggesting personalized diet and exercise plans, which makes self-management much better.

AI systems now analyze lung function tests, images, and symptoms to diagnose COPD accurately and determine how severe it is. Heart patients benefit from wearable sensors that constantly keep track of their condition. AI algorithms can predict cardiac events hours before they happen, giving doctors time to prevent them.

The results speak for themselves. AI chatbots have improved how well patients take their medicine by over 30%, and AI-assisted dosing has cut heart failure hospitalizations by 20%.

What a world of IoMT (Internet of Medical Things) and cloud computing promises. AI doesn’t just react to health crises—it predicts them. These systems spot early warning signs by analyzing tiny changes in vital signs and biomarkers before symptoms show up. This radical alteration moves healthcare from reacting to problems to preventing them.

_{Image Source: MDPI}

Clinical decisions are the lifeblood of effective healthcare. AI now increases physician judgment in ways we’ve never seen before. AI-powered clinical decision support systems (CDSS) lead the vanguard of this transformation and give clinicians analytical insights right when they need them.

AI has become valuable in advancing personalized treatment selection. It analyzes complex datasets, predicts outcomes, and optimizes therapeutic strategies. A study used patients’ gene expression data to train machine learning models that predicted responses to chemotherapy. This is a big deal as it means that the accuracy was above 80% across multiple drugs.

These predictive capabilities work well in mental health treatments, too. Sheu et al.’s research analyzed electronic health records of 17,556 patients to predict responses to different antidepressant classes. The models showed good prediction performance by looking at features that affect treatment selection. This suggests potential for better treatment selection systems.

AI boosts therapy in several ways:

• It adapts cognitive behavioral therapy based on patient progress and unique cognitive patterns
• It analyzes vast datasets of patient histories and treatment responses to find patterns
• It suggests different strategies when patients don’t respond to specific therapy approaches

AI-driven CDSS helps clinicians receive immediate assistance based on patient data. This allows smart decisions about drug selection, dosage, and treatment strategies. CURATE.AI shows this in practice by providing personalized drug dosing. Interviews with 12 oncologists showed they were willing to use the system to optimize doses and boost therapeutic outcomes.

Emergency departments don’t deal very well with overcrowding and resource allocation. AI-based triage systems solve these problems through quick, consistent patient assessment.

AI in triage has brought the most important improvements in predictive accuracy, disease identification, and risk assessment. Machine learning models consistently show better discrimination abilities than conventional triage systems. These models reduced mistreatment rates of critically ill patients to 0.9% compared to 1.2% with traditional systems.

XGBoost, a machine learning algorithm, achieved remarkable results in predicting triage preparation with 83.6% sensitivity, 78.9% specificity, and 80.2% overall prediction accuracy. The Random Forest algorithm showed superior accuracy in determining case severity and predicting triage levels.

AI systems help emergency physicians make time-critical decisions effectively. The KATE™ triage model achieved 75.7% accuracy in predicting Emergency Severity Index acuity assignments. This outperformed triage nurses, who achieved 59.8% – a 26.9% improvement.

Without doubt, these systems make a ground impact. Brussels University Hospital’s AI integration into radiology workflows helped practitioners manage huge volumes of images. The emergency ward uses this technology because it’s fast and simple to use.

AI-enhanced CDSS increases clinical decision-making without replacing human judgment. AI can spot patterns, but providers interpret these signals, add context, and decide the next steps in patient care. This partnership between human expertise and artificial intelligence creates a powerful model to improve healthcare delivery. It combines the best of both worlds to boost patient outcomes.

_{Image Source: Lindy}

Medication errors and adverse reactions rank as the fourth leading cause of death in the United States. These errors cost more than $500 billion each year. AI now offers advanced solutions to optimize drug dosing, monitor therapeutic levels, and predict harmful reactions before patients experience them.

CURATE AI marks a significant advance in medication safety. This AI-powered platform creates tailored dosing profiles from a patient’s small dataset. The technology stands apart from standard methods because it:

• Creates dynamic dose recommendations based on the relationship between drug dose and patient response
• Updates recommendations as new patient data comes in
• Relies solely on individual patient information to ensure fair algorithmic outcomes

Clinical trials have confirmed the platform’s effectiveness across several conditions, including prostate cancer, cognitive therapy, and hypertension. A key study showed CURATE AI successfully optimized doses for 80 patients on combination chemotherapy and improved their outcomes. Twelve oncologists expressed their willingness to use this system to enhance treatment results.

Traditional drug monitoring needs frequent blood sampling and complex modeling. AI now makes concentration predictions more accurate with fewer samples.

XGBoost and similar machine learning models work better than traditional Bayesian estimation methods when measuring residual mean squared error and prediction error. Tacrolimus monitoring shows ML models need just two concentration measurements to outperform standard three-concentration Bayesian estimates.

Neural networks with long short-term memory predict valproic acid levels in older adults more accurately than standard pharmacokinetic models. These improvements mean fewer blood draws while maintaining or improving accuracy.

Quick detection of adverse drug reactions (ADRs) helps establish a drug’s safety profile. ML algorithms analyze large datasets to spot high-risk patients before prescribing medication.

Recent federated learning methods show promise in ADR prediction. They use distributed healthcare data without moving sensitive information between locations. This approach helps solve the challenge of unbalanced data where ADR cases are rare.

ML models accurately predict kidney problems, heart events, and digestive complications from chemotherapy in cancer care. These models achieve AUC values of 0.66 to 0.90 for heart-related events. Predictions for digestive issues like nausea score between 0.81 and 0.85.

Patient age groups and gender rank among the top five factors that predict ADRs. Models become more accurate when they combine demographic information with molecular and biological data.

AI technologies strengthen mental health support and patient engagement by creating tailored therapeutic experiences. State-of-the-art applications now take care beyond traditional clinical settings and tackle critical availability barriers.

Intelligent conversational agents have shown remarkable results in delivering cognitive-behavioral therapy and substance use interventions. Woebot, Wysa, and Youper lead this state-of-the-art field by offering round-the-clock therapeutic support that users rate highly. Users who participated in these platforms saw improvements by a lot in depressive symptoms (d = 0.46) and anxiety (d = 0.57). A study with Woebot for substance use disorder found major reductions in substance use occasions compared to waitlist controls. Most participants recommended the intervention.

These AI chatbots use natural language processing to help users practice evidence-based techniques like cognitive restructuring, behavioral activation, and mindfulness. We reduced treatment barriers by providing anonymous, stigma-free environments that help people who hesitate to seek traditional therapy.

AI excels at detecting emotional states that affect treatment adherence. To cite an instance, OLLIE—an AI system developed by Pleio—analyzes conversational “vibes” to spot emotional obstacles like fear and confusion that often lead to non-adherence. This helps address the American Medical Association’s finding that emotional factors cause half of the top eight medication adherence barriers.

The results speak for themselves. Patients using Vik, an AI chatbot for breast cancer support, showed over 20% improvement in medication compliance. SMS-based AI reminders also boosted medication refill rates by a lot among older patients with chronic conditions.

About 36% of American adults—roughly 80 million people—have simple or below simple health literacy. This gap affects older adults, African Americans (58% with simple or below simple literacy), and Spanish-speaking patients more (74% with less-than-adequate literacy compared to 7% of English speakers).

AI translation models now turn complex medical language into simpler, more available content to bridge this gap. Natural language processing techniques automatically spot and translate “health illiterate” terms using resources like the CDC Plain Language Thesaurus instead of manual rewrites of materials (75% written at high school or college level despite America’s eighth-grade average reading level).

These technologies revolutionize patient-provider relationships by creating more equal access to healthcare knowledge, whatever your educational background or native language.

Healthcare systems face a crucial challenge in implementing AI fairly in various healthcare settings, despite technological advances. Healthcare providers must ensure these powerful tools help all patients equally instead of making existing gaps wider.

Healthcare providers now utilize AI-powered risk stratification to identify patients who need targeted care. Studies show these population stratification algorithms have helped improve blood pressure control in low-income communities. AI systems analyze administrative data to group patients by risk levels. Research indicates the sickest 3.8% of patients use 37% of healthcare resources.

AI can spot housing insecurity and other social factors that standard electronic health records miss when combined with community data. Success depends on measuring initial performance before deployment to spot existing biases.

Algorithmic bias currently results in 17% lower diagnostic accuracy for minority patients. Multilingual AI agents have shown better results with Spanish-speaking patients during screening outreach compared to traditional methods.

These challenges require:

• Training datasets that represent all groups
• Continuous bias monitoring throughout AI lifecycles
• Human supervision of AI systems
• Partnership with affected communities

Standardized ethical guidelines focused on privacy and security help ensure AI becomes a tool for equality rather than amplifying existing healthcare gaps.

Artificial intelligence has reshaped healthcare delivery and changed patient outcomes in 2025. This piece explores how AI technologies help with staff shortages and improve diagnostic accuracy, treatment planning, genomic medicine, remote monitoring, clinical decisions, medication safety, mental health support, and healthcare equity.

AI has made diagnostic workflows better. Radiologists now detect pneumonia with remarkable precision and reduce radiation exposure by 98%. DeepDR algorithms identify diabetic retinopathy with over 94% accuracy. These systems set new standards instead of just making old processes automatic.

Treatment delays have dropped thanks to AI. Microsoft’s InnerEye makes radiotherapy preparation 90% faster. Oncologists can now complete their work in minutes instead of hours. This speed helps patients survive better and feel less anxious during wait times.

AI makes personalized medicine possible by analyzing genetic profiles for custom treatments. Transcriptomic profiling helps find distinct molecular subtypes. This helps children with specific cancer variants avoid harmful radiation treatments.

Remote monitoring systems have changed the game. They analyze data from wearables and sensors to spot subtle health changes early. Traditional methods would miss these patterns. This makes healthcare more preventive than reactive.

Clinical decision support systems with AI help doctors make better choices without replacing their expertise. Machine learning models predict how well different drugs will work with 80% accuracy. Emergency departments use these systems to sort patients better and reduce mistakes with critical cases.

CURATE AI makes medication safer by creating custom dosing profiles from patient data. The system can predict bad drug reactions before giving medicine. This prevents thousands of harmful events every year.

Mental health care is more available now through AI chat systems that provide cognitive-behavioral therapy. These systems show real improvements in depression and anxiety symptoms. Patients taking breast cancer medicine stick to their treatment plan 20% more when AI monitors them.

These advances need to help everyone equally. AI can find at-risk populations who need extra help. We must be careful when using these systems to avoid making healthcare gaps bigger.

Healthcare systems worldwide are using these technologies more. Human expertise works well with AI to improve patient care. Doctors’ wisdom combines with computer precision to make care more personal, quick, and available to patients, whatever their situation.

AI is revolutionizing healthcare delivery by addressing critical staffing shortages while dramatically improving diagnostic accuracy, treatment speed, and patient outcomes across multiple medical specialties.

• Breakthrough diagnostic accuracy: Deep learning systems now detect pneumonia with 100% precision while reducing radiation exposure by 98%, and they identify diabetic retinopathy with 94% accuracy.

• Faster treatment preparation: Microsoft’s InnerEye cuts radiotherapy planning time by 90%, slashing oncology preparation from hours to minutes and boosting survival rates.

• Personalized medicine becomes reality: AI analyzes genetic profiles to identify distinct cancer subtypes, enabling targeted therapies that eliminate harmful treatments like whole-brain radiation for children.

• Predictive care prevents crises: Remote monitoring systems detect patient deterioration patterns before traditional methods, reducing hospital mortality rates through earlier interventions.

• Smarter medication management: Platforms like CURATE.AI build personalized dosing profiles and predict adverse drug reactions before they happen, preventing thousands of harmful events.

• Mental health support becomes accessible: AI chatbots delivering cognitive-behavioral therapy show measurable improvements in depression (46% effect size) and anxiety (57% effect size) symptoms.

The partnership between human clinical expertise and artificial intelligence creates a powerful framework for delivering more personalized, efficient, and equitable healthcare—transforming reactive medicine into truly predictive care that benefits all patients regardless of their circumstances.

Q1. How is AI improving diagnostic accuracy in healthcare? AI is significantly enhancing diagnostic precision across various medical fields. For example, AI-powered systems can detect pneumonia with 100% accuracy while reducing radiation exposure by 98%, and identify diabetic retinopathy with 94% accuracy. These advancements are setting new standards in early disease detection and patient care.

Q2. What impact is AI having on treatment planning and delivery? AI is dramatically reducing treatment delays and improving efficiency. Technologies like Microsoft’s InnerEye cut radiotherapy preparation time by up to 90%, allowing oncologists to complete tasks in minutes rather than hours. This acceleration directly improves patient outcomes by enabling faster interventions and reducing waiting times for critical treatments.

Q3. How does AI contribute to personalized medicine? AI analyzes individual genetic profiles to tailor treatments to specific patients. By identifying distinct molecular subtypes of diseases, AI enables precisely targeted therapies. This is particularly beneficial in oncology, where it can help avoid unnecessary treatments and improve outcomes, especially for pediatric patients.

Q4. What role does AI play in remote patient monitoring? AI-powered remote monitoring systems analyze continuous data streams from wearables and ambient sensors to detect subtle signs of patient deterioration before conventional methods would notice them. This shift towards predictive care helps prevent health crises and reduces hospital mortality rates through earlier interventions.

Q5. How is AI enhancing medication safety and patient engagement? AI platforms like CURATE.AI create personalized drug dosing profiles and predict adverse reactions before medication administration, potentially preventing thousands of harmful events annually. Additionally, AI chatbots delivering cognitive-behavioral therapy show measurable improvements in mental health symptoms, while also increasing medication adherence by over 20% for conditions like breast cancer.