AI is increasingly becoming omnipresent across the entire spectrum of the healthcare industry, including the 5 Ps: payer, provider, policy maker/government, patients, and product manufacturers. AI leverages sophisticated computation and inference to generate insights, enables the system to reason and learn, and empowers clinician decision-making through augmented intelligence.
The field of precision medicine is similarly experiencing rapid growth. The field has since evolved to recognize how the intersection of multi?omic data combined with medical history, social/behavioral determinants, and environmental knowledge precisely characterizes health states, disease states, and therapeutic options for affected individuals.
Precision medicine offers healthcare providers the ability to discover and present information that either validates or alters the trajectory of a medical decision from one that is based on the evidence for the average patient, to one that is based upon an individual’s unique characteristics. Precision medicine methods identify phenotypes of patients with less common responses to treatment or unique healthcare needs. Precision medicine, integrated into healthcare, has the potential to yield more precise diagnoses, predict disease risk before symptoms occur, and design customized treatment plans that maximize safety and efficiency.
Applications and Recent Developments
Genomic Consideration and Therapy Planning
The significance of genomic consideration lies in the fact that patients with pharmacogenomically actionable variants may require altered prescription or dosing. AI has proven efficiency in high throughput genome interpretation. These interpretations identify links among genomic variation and disease presentation, therapeutic success, and prognosis. The Clinical Pharmacogenetics Implementation Consortium published genotype?based drug guidelines to help clinicians optimize drug therapies with genetic test results.
Deep learning has been used to propose 3D protein configurations, identify transcription start sites, model regulatory elements, and predict gene expression from genotype data. AI-mediated analysis has been used in medulloblastoma to administer the right treatment, at the right dosage, to the right cohort of pediatric patients. The treatment required only chemotherapy obviating the need for radiation.
Radiogenomics employs AI in imaging recognition and focuses on establishing an association between cancer imaging features and gene expression to predict a patient’s risk of developing toxicity following radiotherapy. It is primarily evident in radiogenomic associations in breast cancer, liver cancer, and colorectal cancer. Further, AI demonstrates potential applications in knowing the response to a therapy.
Environmental Considerations in Therapy Planning
AI can take care of the availability of expertise in remote locations including the availability of trained professionals at the point of need. AI has in various situations, augmented diagnostic capabilities in resource-poor locations including the identification of patients with malaria and cervical cancer using deep learning, predicting infectious disease outbreaks, environmental toxins exposure, and allergen load.
Therapy Planning
Cardiovascular medicine has a long history of embracing predictive modeling to assess patient risk. It has been recently used to predict heart failure and other serious cardiac events in asymptomatic individuals. As AI approaches excel at discovering complex relationships, they can be successfully applied to study the interplay of various factors including gender, genetics, lifestyle, and environment to obtain heterogeneity of data.
AI enabled recognition of phenotype features through EHR or images and matching those features with genetic variants may allow faster genetic disease diagnosis. Rapid whole?genome sequencing and NLP?enabled automated phenotyping can be used for fast and accurate diagnosis of genetic diseases in ill infants.
Nongenomic Considerations in Risk Prediction
Automated speech analytics can be used in the early diagnosis of dementia, minor cognitive impairment, Parkinson’s disease, and other mental disorders. Efforts also are underway to detect changes in mental health using smartphone sensors. AI-assisted monitoring can assess the risk of intrapartum stress during labor and decide between cesarean section and normal deliveries to reduce perinatal complications and stillbirths.
AI can also detect polyps in colonoscopy. Adoption of AI during endoscopy can lead to accurate detection of benign adenoma with a reduced risk of unwarranted polypectomy at a reduced cost. The use of AI-mediated image analysis is likely to increase for detection of diabetic retinopathy, metastasis in cancer, and benign melanoma. It is also a vital component of the direct-to-consumer diagnostic tool for anemia.
Home monitoring and wearable devices are widely used for monitoring and detecting diabetes, epilepsy, pain management, Parkinson’s disease, cardiovascular disease, sleep disorders, cancer, and obesity. They provide continuous multidimensional measurements of preselected biomarkers to detect minimum residual disease and monitor disease progression. Digital biomarkers are expected to facilitate remote disease monitoring and support decentralized clinical trials.
Challenges to AI in Precision Medicine
Generalization
The efficiency of AI lies in its capability to work accurately in a reliable, safe, and generalizable manner. Currently, the AI models are trained to work on an institutional basis in terms of coding definitions, report formats, or cohort diversity leading to training for one site that does not work well with another site's data, making generalization difficult.
Bias
The health data can be biased in terms of sampling, values, and imputation methods. An AI model trained on such data tends to make unfavorable decisions toward a particular group. This can potentially harm clinical applicability and health quality.
To address this issue many technologies have been developed and the most widely accepted is the one developed by IBM. IBM has developed an online toolkit namely AI Farness 360. It implements a comprehensive set of fairness metrics to examine the bias among dataset models and algorithms to mitigate bias in classifiers.
Socio-Environmental Factors
The environmental factors and workflow impact the performance and clinical efficacy of an AI model. Additionally, travel and associated costs may restrict sample points to participate in a study. This highlights the importance of validation of the AI model in the clinical environment and consideration of the iteration loop before applying the AI system widely.
Data Safety and Privacy
Data privacy is an important individual concern while using AI-enabled services. This emphasizes the need to develop a well-regulated ecosystem for data storage, management, and sharing. This might require further technological development and the development of new regulations and business models.
Studies suggest that translational research exploring this convergence between precision medicine and AI will help solve the most difficult challenges facing precision medicine, especially those in which nongenomic and genomic determinants, combined with information from patient symptoms, clinical history, and lifestyles, will facilitate personalized diagnosis and prognostication. Also, the use of AI in precision medicine is deemed to augment personal medical diagnosis and related therapeutic interventions. This would aid in the early detection and treatment of diseases. Still, more work needs to be done to test, validate, and change treatment practices for the successful integration of AI and precision medicine.