1. A later warning system of malignant tumour identification and patient sentiment tracking based on artificial intelligence that includes a medical imaging acquisition module (configured to receive a diagnostic imaging input of medical devices such as magnetic resonance imaging scanners, computed tomography scanners, mammography equipment, and ultrasound devices) and an image preprocessing module (configured to perform noise reduction, normalization, contrast enhancement, and segmentation of possible tumour appearances), a deep learning tumour detection module (configured to analyse the pre-processed medical images using convolutional neural network architectures to extract hierarchical

2. The system as specified by claim 1 that involves the image preprocessing module further enhancing adaptive filtering and intensity normalization in order to enhance the visibility of suspicious tissue structures in the medical imaging data.

3. The system presented by claim 1 that will be described in this paper where the tumour detection module will be based on deep neural network structures consisting of a convolutional layer, a pooling layer, activation functions as well as classification layers to automatically identify tumour related features in imaging data.

4. The system as per claim 1 in which the tumour classification process generates probabilistic outputs, which are the probability that a detected tumor is of malignant tumour.

5. The system as per claim 1 where the patient sentiment analysis module as used in the system deploys the natural language processing techniques such as tokenization, stop word removal, linguistic normalization, and sentiment polarity classification as used in evaluating emotional states of the patient.

6. The system as per claim 1 where the sentiment analysis module uses machine learning or transformer based language models to identify emotional signs that depict psychological distress or emotional wellbeing.

7. The system in claim 1 where the predictive healthcare analytics engine integrates the imaging based tumour traits, patient sentiment ratings and the past clinical records to come up with the overall patient risk profiles.

8. The system under claim 1 under which the early warning system sends notification events when predetermined diagnostic thresholds related to the likelihood of malignant tumours or emotional distress objectives are surpassed.

9. The system as per claim 1 also including an explainable artificial intelligence module programmed to produce visual interpretation results such as heat maps and highlighted tumour region to help clinicians in interpreting the diagnostic predictions made by the artificial intelligence models.

10. An algorithm defined as a computer implemented approach to the early detection of malignant tumours and monitoring patient sentiment including receiving medical imaging data as acquired by diagnostic imaging devices, preprocessing the images to support diagnostic related features, extraction of tumour related traits using deep learning models, classification of tumour types based on probabilistic prediction algorithms, analysis of patient communication data using natural language processing techniques to determine emotional sentiment indicators, the combination of tumour detection results and sentiment analysis results through a predictive healthcare analytics engine, providing early warning indicators when high risk tumour patterns or emotional distress indicators are identified, and presentation

Description:
The current invention is associated with an AI-based early warning system to detect malignant tumours and monitor patients sentiments, which combines the most recent technology in artificial intelligence, medical image analysis, and natural language processing in order to increase cancer diagnosis and patient monitoring. The system will be created to support the work of the healthcare professionals since it will allow determining the exact location of the tumour and at the same time assessing the emotional state of the diagnosed and treated patients.
The invention proposes an integrated system, which involves the medical imaging diagnosis and the sentiment analysis of a patient to provide predictive information and early warnings. The proposed system architecture as shown in FIG. 1 comprises a number of modules that are
interconnected and under which image acquisition, data preprocessing, tumour detection, sentiment analysis, predictive analytics, and clinical decision support modules are found.
The system has a medical imaging acquisition module (101) that gathers diagnostic images of different medical imaging equipment like magnetic resonance imaging (MRI), computed tomography (CT), mammography scanners, and ultrasound systems. These are imaging devices which give high-resolution medical images with anatomical details of internal tissues and organs. The obtained data can be two-dimensional or three dimensional data of imaging that are applied to detect the presence of abnormal tissue structure and tumour formations.
When the imaging data is received, the images are sent to an image preprocessing and enhancement module (102). The module is used to prepare the medical images to undergo further analysis by the use of a number of image enhancement methods meant to enhance diagnostic accuracy. Preprocessing operations used are noise reduction by filtering methods, contrast enhancement, which brings out important structures, image normalization to bring the intensity levels to the same level, and segmentation operations which isolate regions of interest.
In a single embodiment, the segmentation process employs sophisticated algorithms like threshold-based segmentation, region-growing processes or deep learning-based segmentation model to detect the possible tumour regions in the medical images. These broken segments are then sent to the tumour detection module where they are further studied.
The most crucial analytical aspect of the invention is the deep learning tumour detection module (103). This module uses artificial neural network structures that can be trained on complex characteristics of medical imaging data. Preferred embodiment Convolutional neural networks are applied to extract features in segmented tumour regions which are then extracted in an automated fashion.
A typical neural network architecture is divided into several convolutional layers, activation functions, pooling layers and fully connected classification layers. It is a combination of these layers that analyse patterns of space, change in texture, and structural anomalies in the imaging data. The system derives hierarchical feature representations, which represent low-level features such as edges and gradients, and high level features that represent tumour morphology.
After extracting the features, the classification subsystem appraises extracted features to give a response to whether the detected lesion is benign tumour or malignant tumour. Probabilistic outputs of the classification model are used to show the risk of malignancy. These outcomes are further abridged under validation systems and confidence scoring to make sure that there is diagnostic reliability.
Besides tumour detection, the invention also includes patient sentiment analysis module (104) as such an evaluation deals with emotional responses that are expressed by patients during medical visits and treatment procedures. This module works with the textual information derived due to the use of electronic health records, patient feedback forms, transcripts of online consultations, and telemedicine communication tools.
The sentiment analysis module is the algorithm based on natural language processing to analyze the language pattern of patients and identify emotional indicators. The first stage
involves text preprocessing steps such as tokenization, removal of stop words and linguistic normalization. The analysis of the processed text is further divided into machine learning models that can discover the sentiment polarity and emotional intensity.
In one application, language models using transformers or deep neural networks are used to recognize such emotional conditions as anxiety, stress, fear, sadness or optimism. The sentiment analysis module is an output that takes the form of sentiment scores, which indicate the emotional state of the patient in diagnostic interactions.
The results of the tumour detection module and sentiment analysis module are sent to a predictive healthcare analytics engine (105). It is an engine that merges the imaging-based diagnostics features with sentiment sensors and past clinical records to create a complete patient risk analysis.
The predictive analytics engine uses machine learning algorithms to determine the relationship between tumour features and the emotional conditions of patients. The analysis of such multimodal data sources will enable the system to identify trends that can cause possible health risks or complications. To illustrate, if there is a suspicious tumour feature with a high amount of patient anxiety, this would raise the level of priority alert when it comes to clinical intervention.
It also has a system of early warning generation which generates warnings whenever predefined diagnostic thresholds are breached. These warnings alert medical personnel in case of the identification of potentially malignant tumour patterns or the analysis of the sentiment of the patient who has undergone a serious emotional trauma and needs psychological assistance.
To build trust and transparency in the clinical setting, explainable artificial intelligence methods are implemented in the invention. These methods produce visual descriptions of the areas of medical images that produce diagnostic predictions. Attention maps or heat maps can be created to indicate the position of the tumours in the imaging data.
In a similar manner, the sentiment analysis module can give interpretable results on certain words/phrases in the patient communications that led to a classification of sentiment. This openness will enable clinicians to see the way the artificial intelligence system to its conclusions.
The ultimate products of the system are displayed in the form of a clinical decision support interface (106). This interface gives healthcare experts a full-fledged dashboard which shows diagnostic outcomes, maps of tumour localization, sentiment analysis signals and risk forecast warnings. The interface can also show trend of past patient data so that clinicians can also monitor how the tumour progresses and emotional health change with time.
Clinical decision support interface will have the capability to integrate with the existing hospital information systems and electronic health record systems to ensure smooth data exchange and workflow integration in the healthcare settings.
In general, the current invention offers an innovative and combined solution to the cancer diagnosis and monitoring of patients by merging tumour detection with artificial intelligence
and automatic patient sentiment tracking. This two-fold analysis ability increases the degree of diagnosis, aids in the early diagnosis of malignant tumours and helps the medical professionals to check the psychological wellbeing of the patients during the process of treatment.
The invention can also be seen to make clinical decision support significantly better by using the combination of various artificial intelligence technologies in a single system architecture which has contributed positively to the development of intelligent healthcare systems. , Claims:1. A later warning system of malignant tumour identification and patient sentiment tracking based on artificial intelligence that includes a medical imaging acquisition module (configured to receive a diagnostic imaging input of medical devices such as magnetic resonance imaging scanners, computed tomography scanners, mammography equipment, and ultrasound devices) and an image preprocessing module (configured to perform noise reduction, normalization, contrast enhancement, and segmentation of possible tumour appearances), a deep learning tumour detection module (configured to analyse the pre-processed medical images using convolutional neural network architectures to extract hierarchical
2. The system as specified by claim 1 that involves the image preprocessing module further enhancing adaptive filtering and intensity normalization in order to enhance the visibility of suspicious tissue structures in the medical imaging data.
3. The system presented by claim 1 that will be described in this paper where the tumour detection module will be based on deep neural network structures consisting of a convolutional layer, a pooling layer, activation functions as well as classification layers to automatically identify tumour related features in imaging data.
4. The system as per claim 1 in which the tumour classification process generates probabilistic outputs, which are the probability that a detected tumor is of malignant tumour.
5. The system as per claim 1 where the patient sentiment analysis module as used in the system deploys the natural language processing techniques such as tokenization, stop word removal, linguistic normalization, and sentiment polarity classification as used in evaluating emotional states of the patient.
6. The system as per claim 1 where the sentiment analysis module uses machine learning or transformer based language models to identify emotional signs that depict psychological distress or emotional wellbeing.
7. The system in claim 1 where the predictive healthcare analytics engine integrates the imaging based tumour traits, patient sentiment ratings and the past clinical records to come up with the overall patient risk profiles.
8. The system under claim 1 under which the early warning system sends notification events when predetermined diagnostic thresholds related to the likelihood of malignant tumours or emotional distress objectives are surpassed.
9. The system as per claim 1 also including an explainable artificial intelligence module programmed to produce visual interpretation results such as heat maps and highlighted tumour region to help clinicians in interpreting the diagnostic predictions made by the artificial intelligence models.
10. An algorithm defined as a computer implemented approach to the early detection of malignant tumours and monitoring patient sentiment including receiving medical imaging data as acquired by diagnostic imaging devices, preprocessing the images to support diagnostic related features, extraction of tumour related traits using deep learning models, classification of tumour types based on probabilistic prediction algorithms, analysis of patient communication data using natural language processing techniques to determine emotional sentiment indicators, the combination of tumour detection results and sentiment analysis results through a predictive healthcare analytics engine, providing early warning indicators when high risk tumour patterns or emotional distress indicators are identified, and presentation